54:I[191445,["/_next/static/immutable/chunks/42yzn81wlo94m.js","/_next/static/immutable/chunks/2bm6_f2j__8vm.js","/_next/static/immutable/chunks/3i5397s96cu0a.js","/_next/static/immutable/chunks/2x03jtn4ys_wx.js","/_next/static/immutable/chunks/2f4lnnvm42_vp.js","/_next/static/immutable/chunks/16-tr_iilxrhb.js","/_next/static/immutable/chunks/2888ntn3nn8dg.js","/_next/static/immutable/chunks/2xqpa5c7ogclt.js","/_next/static/immutable/chunks/41cng7f4va9y-.js","/_next/static/immutable/chunks/2r_ddychp2ui5.js","/_next/static/immutable/chunks/1g9smhz0fks9x.js","/_next/static/immutable/chunks/2o71xrq9b0yu6.js","/_next/static/immutable/chunks/3qs9vw68v38q9.js","/_next/static/immutable/chunks/1rm5ppn9o7oew.js","/_next/static/immutable/chunks/2hbgrpnh1fmap.js","/_next/static/immutable/chunks/0tfe231gghziz.js","/_next/static/immutable/chunks/3ol7h3r_z2pit.js","/_next/static/immutable/chunks/3skty2nb9a-2-.js","/_next/static/immutable/chunks/3oekgz5eyywh-.js","/_next/static/immutable/chunks/1e0jyw4aho8bn.js","/_next/static/immutable/chunks/2ri0u_hhdt48v.js","/_next/static/immutable/chunks/25buewscho2w5.js","/_next/static/immutable/chunks/3u4carzbf5fig.js","/_next/static/immutable/chunks/0ewa-4giwctvp.js","/_next/static/immutable/chunks/0fopioox00_wn.js","/_next/static/immutable/chunks/17sbo0v8uw3y7.js","/_next/static/immutable/chunks/05727lidoh_64.js","/_next/static/immutable/chunks/0b0jg9r2u87k8.js","/_next/static/immutable/chunks/459bt34i9w_4u.js","/_next/static/immutable/chunks/2xi3oa1rgz4zi.js","/_next/static/immutable/chunks/0nuoj_60cpek_.js","/_next/static/immutable/chunks/25x_rwsdb74m3.js","/_next/static/immutable/chunks/0p2ndf8a03w3d.js","/_next/static/immutable/chunks/3rpw8gofk-lvz.js","/_next/static/immutable/chunks/0yqqzq4qo78e-.js","/_next/static/immutable/chunks/1lvgxqj71539d.js","/_next/static/immutable/chunks/412rzi0r7ty5b.js","/_next/static/immutable/chunks/2m_mu4_r75f07.js","/_next/static/immutable/chunks/084wa77nfb_vx.js","/_next/static/immutable/chunks/3mt6t4u0f9pd2.js","/_next/static/immutable/chunks/1lgj52kxs241t.js","/_next/static/immutable/chunks/2_9cwii8mx0pv.js","/_next/static/immutable/chunks/4579_u3pt_bl4.js","/_next/static/immutable/chunks/3-y1wg0q2kip1.js","/_next/static/immutable/chunks/3du8tbb5hc0uo.js"],"WorkTabProvider"] 55:I[649892,["/_next/static/immutable/chunks/42yzn81wlo94m.js","/_next/static/immutable/chunks/2bm6_f2j__8vm.js","/_next/static/immutable/chunks/3i5397s96cu0a.js","/_next/static/immutable/chunks/2x03jtn4ys_wx.js","/_next/static/immutable/chunks/2f4lnnvm42_vp.js","/_next/static/immutable/chunks/16-tr_iilxrhb.js","/_next/static/immutable/chunks/2888ntn3nn8dg.js","/_next/static/immutable/chunks/2xqpa5c7ogclt.js","/_next/static/immutable/chunks/41cng7f4va9y-.js","/_next/static/immutable/chunks/2r_ddychp2ui5.js","/_next/static/immutable/chunks/1g9smhz0fks9x.js","/_next/static/immutable/chunks/2o71xrq9b0yu6.js","/_next/static/immutable/chunks/3qs9vw68v38q9.js","/_next/static/immutable/chunks/1rm5ppn9o7oew.js","/_next/static/immutable/chunks/2hbgrpnh1fmap.js","/_next/static/immutable/chunks/0tfe231gghziz.js","/_next/static/immutable/chunks/3ol7h3r_z2pit.js","/_next/static/immutable/chunks/3skty2nb9a-2-.js","/_next/static/immutable/chunks/3oekgz5eyywh-.js","/_next/static/immutable/chunks/1e0jyw4aho8bn.js","/_next/static/immutable/chunks/2ri0u_hhdt48v.js","/_next/static/immutable/chunks/25buewscho2w5.js","/_next/static/immutable/chunks/3u4carzbf5fig.js","/_next/static/immutable/chunks/0ewa-4giwctvp.js","/_next/static/immutable/chunks/0fopioox00_wn.js","/_next/static/immutable/chunks/17sbo0v8uw3y7.js","/_next/static/immutable/chunks/05727lidoh_64.js","/_next/static/immutable/chunks/0b0jg9r2u87k8.js","/_next/static/immutable/chunks/459bt34i9w_4u.js","/_next/static/immutable/chunks/2xi3oa1rgz4zi.js","/_next/static/immutable/chunks/0nuoj_60cpek_.js","/_next/static/immutable/chunks/25x_rwsdb74m3.js","/_next/static/immutable/chunks/0p2ndf8a03w3d.js","/_next/static/immutable/chunks/3rpw8gofk-lvz.js","/_next/static/immutable/chunks/0yqqzq4qo78e-.js","/_next/static/immutable/chunks/1lvgxqj71539d.js","/_next/static/immutable/chunks/412rzi0r7ty5b.js","/_next/static/immutable/chunks/2m_mu4_r75f07.js","/_next/static/immutable/chunks/084wa77nfb_vx.js","/_next/static/immutable/chunks/3mt6t4u0f9pd2.js","/_next/static/immutable/chunks/1lgj52kxs241t.js","/_next/static/immutable/chunks/2_9cwii8mx0pv.js","/_next/static/immutable/chunks/4579_u3pt_bl4.js","/_next/static/immutable/chunks/3-y1wg0q2kip1.js","/_next/static/immutable/chunks/3du8tbb5hc0uo.js"],"WorkHeader"] 56:T1cfc,FoldCraft: An Open Framework for De Novo Design of EGFR-Targeting Nanobodies 1. Overview This project aims to experimentally validate nanobody binders against EGFR, designed entirely from scratch using our AI-powered platform, FoldCraft. FoldCraft uses AlphaFold2-Multimer as a generative engine and has already demonstrated performance superior to state-of-the-art (SOTA) models like RFdiffusion and RFAntibody in in silico benchmarks. To test these designs in real-world settings, we will submit our top 50 nanobody candidates to the Adaptyv Bio binding screening platform, a high-throughput experimental assay that rapidly evaluates the binding ability of synthetic proteins. This work builds on a strong foundation: FoldCraft placed 2nd in Round 1 of Adaptyv Bio’s EGFR binder design competition, showing its potential to generate high-quality miniproteins binders with therapeutic relevance. The project will be completed in 6 months, with no additional infrastructure or ethics approvals required. Most importantly, we commit to open-sourcing all binder sequences and binding data — making this a fully transparent and reusable resource for the broader scientific community. 2. Introduction The Challenge of De Novo Binder Design Designing proteins that can selectively bind to molecular targets — like antibodies that block cancer pathways or neutralize pathogens — is one of the grand challenges in bioengineering. Traditional approaches are often slow, expensive, and dependent on animal immunization or brute-force screening of vast experimental libraries. But what if we could design these proteins directly from code? Recent advances in AI, particularly models like AlphaFold, RFdiffusion, and RFAntibody, have brought us closer to this vision. However, most existing tools still struggle to control the 3D fold of the resulting protein, limiting their ability to produce compact, stable, and therapeutically relevant scaffolds such as VHH nanobodies. This is particularly important when targeting proteins like EGFR (Epidermal Growth Factor Receptor) — a well-characterized cancer target implicated in multiple tumor types. Its clinical relevance makes EGFR an ideal benchmark for evaluating the next generation of AI-driven protein design tools. Enter FoldCraft FoldCraft introduces a new approach called fold-conditioned hallucination, which combines AlphaFold2-Multimer’s generative capabilities with a single, interpretable contact map loss to guide structure formation. Rather than producing arbitrary shapes, FoldCraft steers sequence design toward antibody-like folds, enabling precise control over structural topology from the start. In benchmarking studies, this simple yet powerful method leads to dramatic improvements. FoldCraft has outperformed state-of-the-art models like RFdiffusion and RFAntibody across multiple targets, achieving up to 19.5% success rates in generating valid binders against EGFR, PD-1, and PD-L1 — compared to less than 1% in many competing models. Why This Project Matters We believe FoldCraft’s nanobody designs are ready for experimental testing. This project will determine: · Whether FoldCraft’s in silico success translates to real-world binding. · Whether AlphaFold3 confidence scores predict actual binding. · Whether we can publicly release a dataset of fully de novo nanobody sequences and measured affinities — something rarely done in the field. Key Hypotheses H1: FoldCraft-designed nanobodies will show measurable binding to EGFR in wet-lab screening. H2: AlphaFold3 metrics (ipTM, pLDDT, iPAE) correlate with binding strength. H3: Optimization of only non-interface regions with SolMPNN will improve experimental binding compared to existing strategies. 3. Methods Sample We will generate a large initial set of at least 500 nanobody sequences against EGFR using FoldCraft, focusing on structural designs that adhere to a canonical VHH fold. All sequences will undergo interface-preserving optimization using SolMPNN, and only those passing strict structural filters (e.g., AlphaFold3 plDDT > 70, ipAE 0.5) will be retained. From this filtered pool, we will select the top 50 designs based on AlphaFold3 confidence metrics and sequence/epitope diversity for submission to experimental screening. In addition to high-throughput binding screens, we will order the top-ranked (Top-1) FoldCraft design and perform a functional validation in EGFR-positive cancer cell lines using standard cell-viability assays to assess antiproliferative effects. These experiments will be conducted with existing institutional resources and/or in-kind support without altering the $5,000 allocation to Adaptyv Bio. Procedures1. Step 1: In Silico Design · Generate nanobody scaffolds conditioned on the VHH fold. · Define contact regions based on known EGFR epitopes. · Optimize sequences using SolMPNN. 2. Step 2: Selection and Submission: · Rank designs by AlphaFold3 metrics. · Cluster for diversity and select 50 designs. 3. Step 3: Experimental Screening (Adaptyv Bio): · Submit all 50 sequences to Adaptyv Bio, who will synthesize, express, and test them against EGFR. · Results will include quantitative binding scores. 4. Step 4: Data Analysis: · Compute correlations between AlphaFold3 predictions and experimental scores. · Visualize all structures, docking geometries, and experimental rankings. Ethics and Data Management Cell-line work will follow institutional biosafety policies; authenticated lines and mycoplasma-free status will be documented. All raw viability data, analysis scripts, and the Top-1 sequence will be open-sourced alongside binding data. All sequences, AlphaFold3 models, and binding data will be: · Uploaded to GitHub (FoldCraft repo) · Published on Zenodo with open CC-BY licensing · Included in a future preprint or publication 4. Pilot Data We have already validated FoldCraft computationally and published a preprint: · Designed over 160 EGFR nanobody binders. · Achieved >19% success under AlphaFold3 evaluation thresholds. · Outperformed RFAntibody (SOTA model), which failed to generate binders passing ipAE or ipTM cutoffs on diverse set of therapeutically relevant targets (PD-L1, PD1, EGFR, IFNAR). Additionally, our pipeline placed 2nd out of 74 submissions in the Adaptyv Bio EGFR binder design competition, by designing de novo EGFR miniproteins binder with KD of 4.77*10-6 M. This proposal builds directly on that momentum. 5. Budget All funds will be used to run 50 binding screens at Adaptyv Bio. Item Description Cost (USD) Adaptyv Bio binder screening 50 de novo nanobody sequences vs. EGFR $5000 Total $5000 Computational modeling is performed on existing local GPU infrastructure (RTX 3090/4080/4090). 6. References 1. Rustamov, K.R., and Baev, A.Y. Fold-Conditioned De Novo Binder Design via AlphaFold2-Multimer Hallucination. bioRxiv, 2025. 2. Cotet, T.-S., et al. Crowdsourced Protein Design: Lessons From the Adaptyv EGFR Binder Competition. bioRxiv, 2025. 3. Jumper, J. et al. Highly accurate protein structure prediction with AlphaFold. Nature, 2021. 4. Dauparas, J. et al. Robust deep learning-based protein sequence design using ProteinMPNN. Science, 2022. 5. Bennett, N.R. et al. Atomically accurate de novo design of antibodies with RFdiffusion. bioRxiv, 2025. 6. Abramson, J. et al. Accurate structure prediction of biomolecular interactions with AlphaFold3. Nature, 2024.57:T1e42,FoldCraft: An Open Framework for De Novo Design of EGFR-Targeting Nanobodies 1. Overview This project aims to experimentally validate nanobody binders against EGFR, designed entirely from scratch using our AI-powered platform, FoldCraft. FoldCraft uses AlphaFold2-Multimer as a generative engine and has already demonstrated performance superior to state-of-the-art (SOTA) models like RFdiffusion and RFAntibody in in silico benchmarks. To test these designs in real-world settings, we will submit our top 50 nanobody candidates to the Adaptyv Bio binding screening platform, a high-throughput experimental assay that rapidly evaluates the binding ability of synthetic proteins. This work builds on a strong foundation: FoldCraft placed 2nd in Round 1 of Adaptyv Bio’s EGFR binder design competition, showing its potential to generate high-quality miniproteins binders with therapeutic relevance. The project will be completed in 6 months, with no additional infrastructure or ethics approvals required. Most importantly, we commit to open-sourcing all binder sequences and binding data — making this a fully transparent and reusable resource for the broader scientific community. 2. Introduction The Challenge of De Novo Binder Design Designing proteins that can selectively bind to molecular targets — like antibodies that block cancer pathways or neutralize pathogens — is one of the grand challenges in bioengineering. Traditional approaches are often slow, expensive, and dependent on animal immunization or brute-force screening of vast experimental libraries. But what if we could design these proteins directly from code? Recent advances in AI, particularly models like AlphaFold, RFdiffusion, and RFAntibody, have brought us closer to this vision. However, most existing tools still struggle to control the 3D fold of the resulting protein, limiting their ability to produce compact, stable, and therapeutically relevant scaffolds such as VHH nanobodies. This is particularly important when targeting proteins like EGFR (Epidermal Growth Factor Receptor) — a well-characterized cancer target implicated in multiple tumor types. Its clinical relevance makes EGFR an ideal benchmark for evaluating the next generation of AI-driven protein design tools. Enter FoldCraft FoldCraft introduces a new approach called fold-conditioned hallucination, which combines AlphaFold2-Multimer’s generative capabilities with a single, interpretable contact map loss to guide structure formation. Rather than producing arbitrary shapes, FoldCraft steers sequence design toward antibody-like folds, enabling precise control over structural topology from the start. In benchmarking studies, this simple yet powerful method leads to dramatic improvements. FoldCraft has outperformed state-of-the-art models like RFdiffusion and RFAntibody across multiple targets, achieving up to 19.5% success rates in generating valid binders against EGFR, PD-1, and PD-L1 — compared to less than 1% in many competing models. Why This Project Matters We believe FoldCraft’s nanobody designs are ready for experimental testing. This project will determine: · Whether FoldCraft’s in silico success translates to real-world binding. · Whether AlphaFold3 confidence scores predict actual binding. · Whether we can publicly release a dataset of fully de novo nanobody sequences and measured affinities — something rarely done in the field. Key Hypotheses H1: FoldCraft-designed nanobodies will show measurable binding to EGFR in wet-lab screening. H2: AlphaFold3 metrics (ipTM, pLDDT, iPAE) correlate with binding strength. H3: Optimization of only non-interface regions with SolMPNN will improve experimental binding compared to existing strategies. 3. Methods Sample We will generate a large initial set of at least 500 nanobody sequences against EGFR using FoldCraft, focusing on structural designs that adhere to a canonical VHH fold. All sequences will undergo interface-preserving optimization using SolMPNN, and only those passing strict structural filters (e.g., AlphaFold3 plDDT > 70, ipAE < 10, ipTM > 0.5) will be retained. From this filtered pool, we will select the top 50 designs based on AlphaFold3 confidence metrics and sequence/epitope diversity for submission to experimental screening. In addition to high-throughput binding screens, we will order the top-ranked (Top-1) FoldCraft design and perform a functional validation in EGFR-positive cancer cell lines using standard cell-viability assays to assess antiproliferative effects. These experiments will be conducted with existing institutional resources and/or in-kind support without altering the $5,000 allocation to Adaptyv Bio. Procedures1. Step 1: In Silico Design · Generate nanobody scaffolds conditioned on the VHH fold. · Define contact regions based on known EGFR epitopes. · Optimize sequences using SolMPNN. 2. Step 2: Selection and Submission: · Rank designs by AlphaFold3 metrics. · Cluster for diversity and select 50 designs. 3. Step 3: Experimental Screening (Adaptyv Bio): · Submit all 50 sequences to Adaptyv Bio, who will synthesize, express, and test them against EGFR. · Results will include quantitative binding scores. 4. Step 4: Data Analysis: · Compute correlations between AlphaFold3 predictions and experimental scores. · Visualize all structures, docking geometries, and experimental rankings. Ethics and Data Management Cell-line work will follow institutional biosafety policies; authenticated lines and mycoplasma-free status will be documented. All raw viability data, analysis scripts, and the Top-1 sequence will be open-sourced alongside binding data. All sequences, AlphaFold3 models, and binding data will be: · Uploaded to GitHub (FoldCraft repo) · Published on Zenodo with open CC-BY licensing · Included in a future preprint or publication 4. Pilot Data We have already validated FoldCraft computationally and published a preprint: · Designed over 160 EGFR nanobody binders. · Achieved >19% success under AlphaFold3 evaluation thresholds. · Outperformed RFAntibody (SOTA model), which failed to generate binders passing ipAE or ipTM cutoffs on diverse set of therapeutically relevant targets (PD-L1, PD1, EGFR, IFNAR). Additionally, our pipeline placed 2nd out of 74 submissions in the Adaptyv Bio EGFR binder design competition, by designing de novo EGFR miniproteins binder with KD of 4.77*10-6 M. This proposal builds directly on that momentum. 5. Budget All funds will be used to run 50 binding screens at Adaptyv Bio. Item Description Cost (USD) Adaptyv Bio binder screening 50 de novo nanobody sequences vs. EGFR $5000 Total $5000 Computational modeling is performed on existing local GPU infrastructure (RTX 3090/4080/4090). 6. References 1. Rustamov, K.R., and Baev, A.Y. Fold-Conditioned De Novo Binder Design via AlphaFold2-Multimer Hallucination. bioRxiv, 2025. 2. Cotet, T.-S., et al. Crowdsourced Protein Design: Lessons From the Adaptyv EGFR Binder Competition. bioRxiv, 2025. 3. Jumper, J. et al. Highly accurate protein structure prediction with AlphaFold. Nature, 2021. 4. Dauparas, J. et al. Robust deep learning-based protein sequence design using ProteinMPNN. Science, 2022. 5. Bennett, N.R. et al. Atomically accurate de novo design of antibodies with RFdiffusion. bioRxiv, 2025. 6. Abramson, J. et al. Accurate structure prediction of biomolecular interactions with AlphaFold3. Nature, 2024.58:T7ac3,{"type":"doc","content":[{"type":"heading","attrs":{"id":"5d89198c-b41b-48aa-b57b-b018a9441edb","data-toc-id":"5d89198c-b41b-48aa-b57b-b018a9441edb","textAlign":null,"level":1},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"24px","fontFamily":null,"color":""}},{"type":"bold"}],"text":"FoldCraft: An Open Framework for De Novo Design of EGFR-Targeting Nanobodies"}]},{"type":"heading","attrs":{"id":"8bd7e193-9713-4268-95d2-2bc168848508","data-toc-id":"8bd7e193-9713-4268-95d2-2bc168848508","textAlign":"left","level":2},"content":[{"type":"text","text":"1. Overview"}]},{"type":"paragraph","attrs":{"id":"631f113e-165d-4501-ad12-cb1952f499fd","class":null,"textAlign":"left"},"content":[{"type":"text","text":"This project aims to experimentally validate nanobody binders against EGFR, designed entirely from scratch using our AI-powered platform, FoldCraft. FoldCraft uses AlphaFold2-Multimer as a generative engine and has already demonstrated performance superior to state-of-the-art (SOTA) models like RFdiffusion and RFAntibody in in silico benchmarks."}]},{"type":"paragraph","attrs":{"id":"f488b5b2-fef6-4544-bdee-3f9200b30f75","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","text":"To test these designs in real-world settings, we will submit our top 50 nanobody candidates to the Adaptyv Bio binding screening platform, a high-throughput experimental assay that rapidly evaluates the binding ability of synthetic proteins."}]},{"type":"paragraph","attrs":{"id":"9db0f3ac-d008-4a69-bf73-0d72b2c0f778","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","text":"This work builds on a strong foundation: FoldCraft placed 2nd in Round 1 of Adaptyv Bio’s EGFR binder design competition, showing its potential to generate high-quality miniproteins binders with therapeutic relevance."}]},{"type":"paragraph","attrs":{"id":"b0b93d54-d75f-417d-9d1d-e03354dccc1f","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","text":"The project will be completed in 6 months, with no additional infrastructure or ethics approvals required. Most importantly, we commit to open-sourcing all binder sequences and binding data — making this a fully transparent and reusable resource for the broader scientific community."}]},{"type":"heading","attrs":{"id":"a1b5a040-af3c-4351-ae9b-821d34d2f137","data-toc-id":"a1b5a040-af3c-4351-ae9b-821d34d2f137","textAlign":"left","level":2},"content":[{"type":"text","text":"2. Introduction"}]},{"type":"heading","attrs":{"id":"4b57bbf9-9b28-4e4e-96f2-da87e0392671","data-toc-id":"f72689f6-1861-43e2-a6af-e6e3503dbfd9","textAlign":"left","level":3},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":"","color":""}},{"type":"bold"},{"type":"italic"}],"text":"The Challenge of De Novo Binder Design"}]},{"type":"paragraph","attrs":{"id":"50f9011d-3d9c-4354-827c-194dde66cbec","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":"","color":""}}],"text":"Designing proteins that can selectively bind to molecular targets — like antibodies that block cancer pathways or neutralize pathogens — is one of the grand challenges in bioengineering. Traditional approaches are often slow, expensive, and dependent on animal immunization or brute-force screening of vast experimental libraries."}]},{"type":"paragraph","attrs":{"id":"ad8652c8-d625-4135-93a1-4df9f574ed51","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":"","color":""}}],"text":"But what if we could design these proteins directly from code?"}]},{"type":"paragraph","attrs":{"id":"708a3868-c374-4a2d-a9b1-2423a3a970c3","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":"","color":""}}],"text":"Recent advances in AI, particularly models like AlphaFold, RFdiffusion, and RFAntibody, have brought us closer to this vision. However, most existing tools still struggle to control the 3D fold of the resulting protein, limiting their ability to produce compact, stable, and therapeutically relevant scaffolds such as VHH nanobodies."}]},{"type":"paragraph","attrs":{"id":"2a58d4c3-f574-421e-8040-e8aa8523909e","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":"","color":""}}],"text":"This is particularly important when targeting proteins like EGFR (Epidermal Growth Factor Receptor) — a well-characterized cancer target implicated in multiple tumor types. Its clinical relevance makes EGFR an ideal benchmark for evaluating the next generation of AI-driven protein design tools."}]},{"type":"paragraph","attrs":{"id":"f9af6a7c-38bb-4b30-bfd2-936b6361080d","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":"","color":""}},{"type":"bold"},{"type":"italic"}],"text":"Enter FoldCraft"}]},{"type":"paragraph","attrs":{"id":"7aed97f3-0b46-4f98-a2f0-c42e957467a3","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":"","color":""}}],"text":"FoldCraft introduces a new approach called fold-conditioned hallucination, which combines AlphaFold2-Multimer’s generative capabilities with a single, interpretable contact map loss to guide structure formation. Rather than producing arbitrary shapes, FoldCraft steers sequence design toward antibody-like folds, enabling precise control over structural topology from the start."}]},{"type":"paragraph","attrs":{"id":"1f063c3d-b3db-432d-ad5c-5d92860a7018","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":"","color":""}}],"text":"In benchmarking studies, this simple yet powerful method leads to dramatic improvements. FoldCraft has outperformed state-of-the-art models like RFdiffusion and RFAntibody across multiple targets, achieving up to 19.5% success rates in generating valid binders against EGFR, PD-1, and PD-L1 — compared to less than 1% in many competing models."}]},{"type":"paragraph","attrs":{"id":"993bf57e-3f6d-4624-992e-6d7147442aa2","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":"","color":""}},{"type":"bold"},{"type":"italic"}],"text":"Why This Project Matters"}]},{"type":"paragraph","attrs":{"id":"6da1f3d1-8541-4bd0-a5aa-07719b7a64ff","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":"","color":""}}],"text":"We believe FoldCraft’s nanobody designs are ready for experimental testing. This project will determine:"}]},{"type":"paragraph","attrs":{"id":"50d80d9c-4f3e-48ce-8745-273408dc7d41","class":"MsoListParagraphCxSpFirst","textAlign":"justify"},"content":[{"type":"text","text":"·"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"7pt","fontFamily":"","color":""}}],"text":" "},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":"","color":""}}],"text":"Whether FoldCraft’s in silico success translates to real-world binding."}]},{"type":"paragraph","attrs":{"id":"9da31fda-e494-41ed-a7b3-414bc1f953f7","class":"MsoListParagraphCxSpMiddle","textAlign":"justify"},"content":[{"type":"text","text":"·"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"7pt","fontFamily":"","color":""}}],"text":" "},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":"","color":""}}],"text":"Whether AlphaFold3 confidence scores predict actual binding."}]},{"type":"paragraph","attrs":{"id":"c76369f6-4a07-483b-8413-3c698576bf23","class":"MsoListParagraphCxSpLast","textAlign":"justify"},"content":[{"type":"text","text":"·"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"7pt","fontFamily":"","color":""}}],"text":" "},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":"","color":""}}],"text":"Whether we can publicly release a dataset of fully de novo nanobody sequences and measured affinities — something rarely done in the field."}]},{"type":"paragraph","attrs":{"id":"501d9b20-b52a-45f6-8ba9-3034160509a1","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":"","color":""}},{"type":"bold"},{"type":"italic"}],"text":"Key Hypotheses"}]},{"type":"paragraph","attrs":{"id":"9d9ab480-6793-4a2b-acc1-dba8e97a9e1f","class":"MsoListParagraphCxSpFirst","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":"","color":""}}],"text":"H1: FoldCraft-designed nanobodies will show measurable binding to EGFR in wet-lab screening."}]},{"type":"paragraph","attrs":{"id":"7a2d2d37-7f65-408c-9bbf-d5684eb36c3a","class":"MsoListParagraphCxSpMiddle","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":"","color":""}}],"text":"H2: AlphaFold3 metrics (ipTM, pLDDT, iPAE) correlate with binding strength."}]},{"type":"paragraph","attrs":{"id":"9693f79f-a668-4739-9b6b-c340308750b4","class":"MsoListParagraphCxSpLast","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":"","color":""}}],"text":"H3: Optimization of only non-interface regions with SolMPNN will improve experimental binding compared to existing strategies."}]},{"type":"heading","attrs":{"id":"275805e1-4c5d-41c8-84e4-81496c76c950","data-toc-id":"275805e1-4c5d-41c8-84e4-81496c76c950","textAlign":"left","level":2},"content":[{"type":"text","text":"3. Methods"}]},{"type":"heading","attrs":{"id":"b1856b3e-8765-464b-811c-7006530ed0ef","data-toc-id":"b1856b3e-8765-464b-811c-7006530ed0ef","textAlign":"left","level":3},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}},{"type":"bold"},{"type":"italic"}],"text":"Sample"}]},{"type":"paragraph","attrs":{"id":"031c2453-affd-4b2e-bf7f-723253296c45","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"We will generate a large initial set of at least 500 nanobody sequences against EGFR using FoldCraft, focusing on structural designs that adhere to a canonical VHH fold. All sequences will undergo interface-preserving optimization using SolMPNN, and only those passing strict structural filters (e.g., AlphaFold3 plDDT > 70, ipAE < 10, ipTM > 0.5) will be retained."}]},{"type":"paragraph","attrs":{"id":"6cfd8dd5-bc3a-488d-859f-0059cde1a1d4","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"From this filtered pool, we will select the top 50 designs based on AlphaFold3 confidence metrics and sequence/epitope diversity for submission to experimental screening."}]},{"type":"paragraph","attrs":{"id":"29d34141-58da-4737-a57e-9a0fcd9cc368","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"In addition to high-throughput binding screens, we will "},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}},{"type":"bold"}],"text":"order the top-ranked (Top-1) FoldCraft design"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":" and perform a "},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}},{"type":"bold"}],"text":"functional validation in EGFR-positive cancer cell lines"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":" using standard cell-viability assays to assess antiproliferative effects. These experiments will be conducted with existing institutional resources and/or in-kind support without altering the $5,000 allocation to Adaptyv Bio."}]},{"type":"imageBlock","attrs":{"src":"https://storage.prod.researchhub.com/uploads/notes/users/180556/e37c4ccb-adb0-4c64-bad0-e6b723f992fd/fig1-1.png","width":"100%","align":"center"}},{"type":"paragraph","attrs":{"id":"4302150e-9e32-4bba-9abc-55174fdd8af5","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}},{"type":"bold"},{"type":"italic"}],"text":"Procedures"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"1. Step 1: In Silico Design"}]},{"type":"paragraph","attrs":{"id":"2f275a2d-b813-422b-99b1-82e81cf76f51","class":"MsoListParagraphCxSpFirst","textAlign":"justify"},"content":[{"type":"text","text":"·"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"7pt","fontFamily":null,"color":""}}],"text":" "},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"Generate nanobody scaffolds conditioned on the VHH fold."}]},{"type":"paragraph","attrs":{"id":"adc80019-b817-4d77-9d4f-ea1fac9a4ec1","class":"MsoListParagraphCxSpMiddle","textAlign":"justify"},"content":[{"type":"text","text":"·"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"7pt","fontFamily":null,"color":""}}],"text":" "},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"Define contact regions based on known EGFR epitopes."}]},{"type":"paragraph","attrs":{"id":"918b5f65-4cf8-412a-a276-38b585835186","class":"MsoListParagraphCxSpLast","textAlign":"justify"},"content":[{"type":"text","text":"·"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"7pt","fontFamily":null,"color":""}}],"text":" "},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"Optimize sequences using SolMPNN."}]},{"type":"paragraph","attrs":{"id":"3a391b3c-39b3-4ef0-8feb-7b428aba9bda","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"2. Step 2: Selection and Submission:"}]},{"type":"paragraph","attrs":{"id":"99851ea7-538a-4e85-8f32-65f07c1b537b","class":"MsoListParagraphCxSpFirst","textAlign":"justify"},"content":[{"type":"text","text":"·"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"7pt","fontFamily":null,"color":""}}],"text":" "},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"Rank designs by AlphaFold3 metrics."}]},{"type":"paragraph","attrs":{"id":"1fe18a82-292b-44cd-aabd-b84bf96208c4","class":"MsoListParagraphCxSpLast","textAlign":"justify"},"content":[{"type":"text","text":"·"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"7pt","fontFamily":null,"color":""}}],"text":" "},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"Cluster for diversity and select 50 designs."}]},{"type":"paragraph","attrs":{"id":"b695e9a5-ea63-4f14-b4a1-5c0a32c7d84e","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"3. Step 3: Experimental Screening (Adaptyv Bio):"}]},{"type":"paragraph","attrs":{"id":"4bbe5497-e226-480f-a238-86ecda2b9122","class":"MsoListParagraphCxSpFirst","textAlign":"justify"},"content":[{"type":"text","text":"·"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"7pt","fontFamily":null,"color":""}}],"text":" "},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"Submit all 50 sequences to Adaptyv Bio, who will synthesize, express, and test them against EGFR."}]},{"type":"paragraph","attrs":{"id":"0577c75c-fa64-45d6-b72e-bfb1a31ed4f7","class":"MsoListParagraphCxSpLast","textAlign":"justify"},"content":[{"type":"text","text":"·"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"7pt","fontFamily":null,"color":""}}],"text":" "},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"Results will include quantitative binding scores."}]},{"type":"paragraph","attrs":{"id":"b567fcb0-42c7-47a0-b247-62b0fc2f4695","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"4. Step 4: Data Analysis:"}]},{"type":"paragraph","attrs":{"id":"e7f421ed-4b5c-436b-bfb0-17acd340e5e1","class":"MsoListParagraphCxSpFirst","textAlign":"justify"},"content":[{"type":"text","text":"·"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"7pt","fontFamily":null,"color":""}}],"text":" "},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"Compute correlations between AlphaFold3 predictions and experimental scores."}]},{"type":"paragraph","attrs":{"id":"5a95cf55-9145-43b9-aff4-c952e5a0b3bc","class":"MsoListParagraphCxSpLast","textAlign":"justify"},"content":[{"type":"text","text":"·"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"7pt","fontFamily":null,"color":""}}],"text":" "},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"Visualize all structures, docking geometries, and experimental rankings."}]},{"type":"paragraph","attrs":{"id":"b17aaee2-1499-41b8-9677-149826e265dc","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}},{"type":"bold"}],"text":"Ethics and Data Management"}]},{"type":"paragraph","attrs":{"id":"63f731ba-42c1-4a77-995e-ffc6bc2a2e61","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"Cell-line work will follow institutional biosafety policies; authenticated lines and mycoplasma-free status will be documented. All raw viability data, analysis scripts, and the Top-1 sequence will be "},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}},{"type":"bold"}],"text":"open-sourced"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":" alongside binding data. All sequences, AlphaFold3 models, and binding data will be:"}]},{"type":"paragraph","attrs":{"id":"144dbb26-9c65-42a8-bdd0-5194c0032f45","class":"MsoListParagraphCxSpFirst","textAlign":"justify"},"content":[{"type":"text","text":"·"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"7pt","fontFamily":null,"color":""}}],"text":" "},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"Uploaded to GitHub (FoldCraft repo)"}]},{"type":"paragraph","attrs":{"id":"8a545aff-c72e-4f98-ac3a-d72b4d49b675","class":"MsoListParagraphCxSpMiddle","textAlign":"justify"},"content":[{"type":"text","text":"·"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"7pt","fontFamily":null,"color":""}}],"text":" "},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"Published on Zenodo with open CC-BY licensing"}]},{"type":"paragraph","attrs":{"id":"2a2f17e3-cdcd-4852-8e14-e6c4119207d0","class":"MsoListParagraphCxSpLast","textAlign":"justify"},"content":[{"type":"text","text":"·"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"7pt","fontFamily":null,"color":""}}],"text":" "},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"Included in a future preprint or publication"}]},{"type":"heading","attrs":{"id":"50236b91-422b-493d-aef1-c1e25b4f48c4","data-toc-id":"50236b91-422b-493d-aef1-c1e25b4f48c4","textAlign":"left","level":2},"content":[{"type":"text","text":"4. Pilot Data"}]},{"type":"paragraph","attrs":{"id":"f6254412-d5d0-4490-b914-1cc54c3090d8","class":null,"textAlign":"left"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"We have already validated FoldCraft computationally and published a preprint:"}]},{"type":"paragraph","attrs":{"id":"3b3fae7f-8763-4e01-a4dc-5c108dbc7311","class":"MsoListParagraphCxSpFirst","textAlign":"justify"},"content":[{"type":"text","text":"·"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"7pt","fontFamily":null,"color":""}}],"text":" "},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"Designed over 160 EGFR nanobody binders."}]},{"type":"paragraph","attrs":{"id":"8407bf4e-a76a-4cc9-81a5-746b5b141b62","class":"MsoListParagraphCxSpMiddle","textAlign":"justify"},"content":[{"type":"text","text":"·"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"7pt","fontFamily":null,"color":""}}],"text":" "},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"Achieved >19% success under AlphaFold3 evaluation thresholds."}]},{"type":"paragraph","attrs":{"id":"38409ec6-22d5-4e94-ac1c-363ecc0c9b08","class":"MsoListParagraphCxSpLast","textAlign":"justify"},"content":[{"type":"text","text":"·"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"7pt","fontFamily":null,"color":""}}],"text":" "},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"Outperformed RFAntibody (SOTA model), which failed to generate binders passing ipAE or ipTM cutoffs on diverse set of therapeutically relevant targets (PD-L1, PD1, EGFR, IFNAR)."}]},{"type":"imageBlock","attrs":{"src":"https://storage.prod.researchhub.com/uploads/notes/users/180556/a90c31ee-81d5-473c-aa21-387724163b74/fig4-1.png","width":"100%","align":"center"}},{"type":"paragraph","attrs":{"id":"b55337dd-0682-415b-8b77-96fdcd0962a4","class":"MsoListParagraphCxSpLast","textAlign":"justify"}},{"type":"paragraph","attrs":{"id":"ccf23fd1-2424-4666-a265-8021691af9f3","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"Additionally, our pipeline placed 2nd out of 74 submissions in the Adaptyv Bio EGFR binder design competition, by designing de novo EGFR miniproteins binder with KD of 4.77*10-6 M."}]},{"type":"paragraph","attrs":{"id":"037f1b20-47cb-42f2-98e8-44161e889af6","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"This proposal builds directly on that momentum."}]},{"type":"heading","attrs":{"id":"fefdd03b-f00d-4400-9bc7-f5fa34fb09ca","data-toc-id":"fefdd03b-f00d-4400-9bc7-f5fa34fb09ca","textAlign":"left","level":2},"content":[{"type":"text","text":"5. Budget"}]},{"type":"paragraph","attrs":{"id":"0d759746-fb7d-4c0e-8f2f-16ccdc832a41","class":null,"textAlign":"left"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"All funds will be used to run 50 binding screens at Adaptyv Bio."}]},{"type":"table","attrs":{"id":"4b7f3263-0950-42d2-a36d-d68526a08941"},"content":[{"type":"tableRow","content":[{"type":"tableCell","attrs":{"colspan":1,"rowspan":1,"colwidth":null,"style":"width:2.0in;border:solid windowtext 1.0pt;\n mso-border-alt:solid windowtext .5pt;padding:0in 5.4pt 0in 5.4pt"},"content":[{"type":"paragraph","attrs":{"id":"14feabc7-5b65-4a97-82f7-562f15f2ba53","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"bold"}],"text":"Item"}]}]},{"type":"tableCell","attrs":{"colspan":1,"rowspan":1,"colwidth":null,"style":"width:2.0in;border:solid windowtext 1.0pt;\n border-left:none;mso-border-left-alt:solid windowtext .5pt;mso-border-alt:\n solid windowtext .5pt;padding:0in 5.4pt 0in 5.4pt"},"content":[{"type":"paragraph","attrs":{"id":"75267a8f-ca16-444f-9441-a89cdd71818e","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"bold"}],"text":"Description"}]}]},{"type":"tableCell","attrs":{"colspan":1,"rowspan":1,"colwidth":null,"style":"width:2.0in;border:solid windowtext 1.0pt;\n border-left:none;mso-border-left-alt:solid windowtext .5pt;mso-border-alt:\n solid windowtext .5pt;padding:0in 5.4pt 0in 5.4pt"},"content":[{"type":"paragraph","attrs":{"id":"97c61cf1-9e57-4741-a1fa-73f9747ea1b0","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"bold"}],"text":"Cost (USD)"}]}]}]},{"type":"tableRow","content":[{"type":"tableCell","attrs":{"colspan":1,"rowspan":1,"colwidth":null,"style":"width:2.0in;border:solid windowtext 1.0pt;\n border-top:none;mso-border-top-alt:solid windowtext .5pt;mso-border-alt:solid windowtext .5pt;\n padding:0in 5.4pt 0in 5.4pt"},"content":[{"type":"paragraph","attrs":{"id":"b0e73e55-bdd2-457c-91c0-fc227b7b35a0","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","text":"Adaptyv Bio binder screening"}]}]},{"type":"tableCell","attrs":{"colspan":1,"rowspan":1,"colwidth":null,"style":"width:2.0in;border-top:none;border-left:none;\n border-bottom:solid windowtext 1.0pt;border-right:solid windowtext 1.0pt;\n mso-border-top-alt:solid windowtext .5pt;mso-border-left-alt:solid windowtext .5pt;\n mso-border-alt:solid windowtext .5pt;padding:0in 5.4pt 0in 5.4pt"},"content":[{"type":"paragraph","attrs":{"id":"5c0b13fc-7325-47b3-b703-7c9c4a7e34f9","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","text":"50 de novo nanobody sequences vs. EGFR"}]}]},{"type":"tableCell","attrs":{"colspan":1,"rowspan":1,"colwidth":null,"style":"width:2.0in;border-top:none;border-left:none;\n border-bottom:solid windowtext 1.0pt;border-right:solid windowtext 1.0pt;\n mso-border-top-alt:solid windowtext .5pt;mso-border-left-alt:solid windowtext .5pt;\n mso-border-alt:solid windowtext .5pt;padding:0in 5.4pt 0in 5.4pt"},"content":[{"type":"paragraph","attrs":{"id":"7ddefae1-e326-4e0b-aaa5-14c7a0b8e28e","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","text":"$5000"}]}]}]},{"type":"tableRow","content":[{"type":"tableCell","attrs":{"colspan":1,"rowspan":1,"colwidth":null,"style":"width:2.0in;border:solid windowtext 1.0pt;\n border-top:none;mso-border-top-alt:solid windowtext .5pt;mso-border-alt:solid windowtext .5pt;\n padding:0in 5.4pt 0in 5.4pt"},"content":[{"type":"paragraph","attrs":{"id":"6776658e-9b1c-4513-b430-f6111d8fecde","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"bold"}],"text":"Total"}]}]},{"type":"tableCell","attrs":{"colspan":1,"rowspan":1,"colwidth":null,"style":"width:2.0in;border-top:none;border-left:none;\n border-bottom:solid windowtext 1.0pt;border-right:solid windowtext 1.0pt;\n mso-border-top-alt:solid windowtext .5pt;mso-border-left-alt:solid windowtext .5pt;\n mso-border-alt:solid windowtext .5pt;padding:0in 5.4pt 0in 5.4pt"},"content":[{"type":"paragraph","attrs":{"id":"13c2853c-0ae8-4a7b-8891-8f96d4fa2f98","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"bold"}],"text":" "}]}]},{"type":"tableCell","attrs":{"colspan":1,"rowspan":1,"colwidth":null,"style":"width:2.0in;border-top:none;border-left:none;\n border-bottom:solid windowtext 1.0pt;border-right:solid windowtext 1.0pt;\n mso-border-top-alt:solid windowtext .5pt;mso-border-left-alt:solid windowtext .5pt;\n mso-border-alt:solid windowtext .5pt;padding:0in 5.4pt 0in 5.4pt"},"content":[{"type":"paragraph","attrs":{"id":"4ca8f240-6de8-42cf-aadd-7d867588369c","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"bold"}],"text":"$5000"}]}]}]}]},{"type":"paragraph","attrs":{"id":"33e7af2f-31a6-4b0b-a631-03c0a1eea96e","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"Computational modeling is performed on existing local GPU infrastructure (RTX 3090/4080/4090)."}]},{"type":"heading","attrs":{"id":"771a83fc-c1ab-45d8-bf50-9e87602cd9cf","data-toc-id":"771a83fc-c1ab-45d8-bf50-9e87602cd9cf","textAlign":"left","level":2},"content":[{"type":"text","text":"6. References"}]},{"type":"paragraph","attrs":{"id":"ed1998c9-6268-4ff7-89dd-b7bf9dcb2259","class":null,"textAlign":"left"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"1. Rustamov, K.R., and Baev, A.Y. Fold-Conditioned De Novo Binder Design via AlphaFold2-Multimer Hallucination. bioRxiv, 2025."}]},{"type":"paragraph","attrs":{"id":"ac7518dd-04d1-4a51-abf1-875fb762fd5e","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"2. Cotet, T.-S., et al. Crowdsourced Protein Design: Lessons From the Adaptyv EGFR Binder Competition. bioRxiv, 2025."}]},{"type":"paragraph","attrs":{"id":"a2a04e1a-7936-4180-a737-381dd3fdf1ca","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"3. Jumper, J. et al. Highly accurate protein structure prediction with AlphaFold. Nature, 2021."}]},{"type":"paragraph","attrs":{"id":"2741a462-92e7-4b0f-806c-4f48d1880cbd","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"4. Dauparas, J. et al. Robust deep learning-based protein sequence design using ProteinMPNN. Science, 2022."}]},{"type":"paragraph","attrs":{"id":"bab49267-2936-4518-a055-791cb98f634d","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"5. Bennett, N.R. et al. Atomically accurate de novo design of antibodies with RFdiffusion. bioRxiv, 2025."}]},{"type":"paragraph","attrs":{"id":"bf1f8157-9c12-4b8e-9eb2-d5d31a6d026c","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"6. Abramson, J. et al. Accurate structure prediction of biomolecular interactions with AlphaFold3. Nature, 2024."}]}]}59:T1e42,FoldCraft: An Open Framework for De Novo Design of EGFR-Targeting Nanobodies 1. Overview This project aims to experimentally validate nanobody binders against EGFR, designed entirely from scratch using our AI-powered platform, FoldCraft. FoldCraft uses AlphaFold2-Multimer as a generative engine and has already demonstrated performance superior to state-of-the-art (SOTA) models like RFdiffusion and RFAntibody in in silico benchmarks. To test these designs in real-world settings, we will submit our top 50 nanobody candidates to the Adaptyv Bio binding screening platform, a high-throughput experimental assay that rapidly evaluates the binding ability of synthetic proteins. This work builds on a strong foundation: FoldCraft placed 2nd in Round 1 of Adaptyv Bio’s EGFR binder design competition, showing its potential to generate high-quality miniproteins binders with therapeutic relevance. The project will be completed in 6 months, with no additional infrastructure or ethics approvals required. Most importantly, we commit to open-sourcing all binder sequences and binding data — making this a fully transparent and reusable resource for the broader scientific community. 2. Introduction The Challenge of De Novo Binder Design Designing proteins that can selectively bind to molecular targets — like antibodies that block cancer pathways or neutralize pathogens — is one of the grand challenges in bioengineering. Traditional approaches are often slow, expensive, and dependent on animal immunization or brute-force screening of vast experimental libraries. But what if we could design these proteins directly from code? Recent advances in AI, particularly models like AlphaFold, RFdiffusion, and RFAntibody, have brought us closer to this vision. However, most existing tools still struggle to control the 3D fold of the resulting protein, limiting their ability to produce compact, stable, and therapeutically relevant scaffolds such as VHH nanobodies. This is particularly important when targeting proteins like EGFR (Epidermal Growth Factor Receptor) — a well-characterized cancer target implicated in multiple tumor types. Its clinical relevance makes EGFR an ideal benchmark for evaluating the next generation of AI-driven protein design tools. Enter FoldCraft FoldCraft introduces a new approach called fold-conditioned hallucination, which combines AlphaFold2-Multimer’s generative capabilities with a single, interpretable contact map loss to guide structure formation. Rather than producing arbitrary shapes, FoldCraft steers sequence design toward antibody-like folds, enabling precise control over structural topology from the start. In benchmarking studies, this simple yet powerful method leads to dramatic improvements. FoldCraft has outperformed state-of-the-art models like RFdiffusion and RFAntibody across multiple targets, achieving up to 19.5% success rates in generating valid binders against EGFR, PD-1, and PD-L1 — compared to less than 1% in many competing models. Why This Project Matters We believe FoldCraft’s nanobody designs are ready for experimental testing. This project will determine: · Whether FoldCraft’s in silico success translates to real-world binding. · Whether AlphaFold3 confidence scores predict actual binding. · Whether we can publicly release a dataset of fully de novo nanobody sequences and measured affinities — something rarely done in the field. Key Hypotheses H1: FoldCraft-designed nanobodies will show measurable binding to EGFR in wet-lab screening. H2: AlphaFold3 metrics (ipTM, pLDDT, iPAE) correlate with binding strength. H3: Optimization of only non-interface regions with SolMPNN will improve experimental binding compared to existing strategies. 3. Methods Sample We will generate a large initial set of at least 500 nanobody sequences against EGFR using FoldCraft, focusing on structural designs that adhere to a canonical VHH fold. All sequences will undergo interface-preserving optimization using SolMPNN, and only those passing strict structural filters (e.g., AlphaFold3 plDDT > 70, ipAE < 10, ipTM > 0.5) will be retained. From this filtered pool, we will select the top 50 designs based on AlphaFold3 confidence metrics and sequence/epitope diversity for submission to experimental screening. In addition to high-throughput binding screens, we will order the top-ranked (Top-1) FoldCraft design and perform a functional validation in EGFR-positive cancer cell lines using standard cell-viability assays to assess antiproliferative effects. These experiments will be conducted with existing institutional resources and/or in-kind support without altering the $5,000 allocation to Adaptyv Bio. Procedures1. Step 1: In Silico Design · Generate nanobody scaffolds conditioned on the VHH fold. · Define contact regions based on known EGFR epitopes. · Optimize sequences using SolMPNN. 2. Step 2: Selection and Submission: · Rank designs by AlphaFold3 metrics. · Cluster for diversity and select 50 designs. 3. Step 3: Experimental Screening (Adaptyv Bio): · Submit all 50 sequences to Adaptyv Bio, who will synthesize, express, and test them against EGFR. · Results will include quantitative binding scores. 4. Step 4: Data Analysis: · Compute correlations between AlphaFold3 predictions and experimental scores. · Visualize all structures, docking geometries, and experimental rankings. Ethics and Data Management Cell-line work will follow institutional biosafety policies; authenticated lines and mycoplasma-free status will be documented. All raw viability data, analysis scripts, and the Top-1 sequence will be open-sourced alongside binding data. All sequences, AlphaFold3 models, and binding data will be: · Uploaded to GitHub (FoldCraft repo) · Published on Zenodo with open CC-BY licensing · Included in a future preprint or publication 4. Pilot Data We have already validated FoldCraft computationally and published a preprint: · Designed over 160 EGFR nanobody binders. · Achieved >19% success under AlphaFold3 evaluation thresholds. · Outperformed RFAntibody (SOTA model), which failed to generate binders passing ipAE or ipTM cutoffs on diverse set of therapeutically relevant targets (PD-L1, PD1, EGFR, IFNAR). Additionally, our pipeline placed 2nd out of 74 submissions in the Adaptyv Bio EGFR binder design competition, by designing de novo EGFR miniproteins binder with KD of 4.77*10-6 M. This proposal builds directly on that momentum. 5. Budget All funds will be used to run 50 binding screens at Adaptyv Bio. Item Description Cost (USD) Adaptyv Bio binder screening 50 de novo nanobody sequences vs. EGFR $5000 Total $5000 Computational modeling is performed on existing local GPU infrastructure (RTX 3090/4080/4090). 6. References 1. Rustamov, K.R., and Baev, A.Y. Fold-Conditioned De Novo Binder Design via AlphaFold2-Multimer Hallucination. bioRxiv, 2025. 2. Cotet, T.-S., et al. Crowdsourced Protein Design: Lessons From the Adaptyv EGFR Binder Competition. bioRxiv, 2025. 3. Jumper, J. et al. Highly accurate protein structure prediction with AlphaFold. Nature, 2021. 4. Dauparas, J. et al. Robust deep learning-based protein sequence design using ProteinMPNN. Science, 2022. 5. Bennett, N.R. et al. Atomically accurate de novo design of antibodies with RFdiffusion. bioRxiv, 2025. 6. Abramson, J. et al. Accurate structure prediction of biomolecular interactions with AlphaFold3. Nature, 2024.5a:T7ac3,{"type":"doc","content":[{"type":"heading","attrs":{"id":"5d89198c-b41b-48aa-b57b-b018a9441edb","data-toc-id":"5d89198c-b41b-48aa-b57b-b018a9441edb","textAlign":null,"level":1},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"24px","fontFamily":null,"color":""}},{"type":"bold"}],"text":"FoldCraft: An Open Framework for De Novo Design of EGFR-Targeting Nanobodies"}]},{"type":"heading","attrs":{"id":"8bd7e193-9713-4268-95d2-2bc168848508","data-toc-id":"8bd7e193-9713-4268-95d2-2bc168848508","textAlign":"left","level":2},"content":[{"type":"text","text":"1. Overview"}]},{"type":"paragraph","attrs":{"id":"631f113e-165d-4501-ad12-cb1952f499fd","class":null,"textAlign":"left"},"content":[{"type":"text","text":"This project aims to experimentally validate nanobody binders against EGFR, designed entirely from scratch using our AI-powered platform, FoldCraft. FoldCraft uses AlphaFold2-Multimer as a generative engine and has already demonstrated performance superior to state-of-the-art (SOTA) models like RFdiffusion and RFAntibody in in silico benchmarks."}]},{"type":"paragraph","attrs":{"id":"f488b5b2-fef6-4544-bdee-3f9200b30f75","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","text":"To test these designs in real-world settings, we will submit our top 50 nanobody candidates to the Adaptyv Bio binding screening platform, a high-throughput experimental assay that rapidly evaluates the binding ability of synthetic proteins."}]},{"type":"paragraph","attrs":{"id":"9db0f3ac-d008-4a69-bf73-0d72b2c0f778","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","text":"This work builds on a strong foundation: FoldCraft placed 2nd in Round 1 of Adaptyv Bio’s EGFR binder design competition, showing its potential to generate high-quality miniproteins binders with therapeutic relevance."}]},{"type":"paragraph","attrs":{"id":"b0b93d54-d75f-417d-9d1d-e03354dccc1f","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","text":"The project will be completed in 6 months, with no additional infrastructure or ethics approvals required. Most importantly, we commit to open-sourcing all binder sequences and binding data — making this a fully transparent and reusable resource for the broader scientific community."}]},{"type":"heading","attrs":{"id":"a1b5a040-af3c-4351-ae9b-821d34d2f137","data-toc-id":"a1b5a040-af3c-4351-ae9b-821d34d2f137","textAlign":"left","level":2},"content":[{"type":"text","text":"2. Introduction"}]},{"type":"heading","attrs":{"id":"4b57bbf9-9b28-4e4e-96f2-da87e0392671","data-toc-id":"f72689f6-1861-43e2-a6af-e6e3503dbfd9","textAlign":"left","level":3},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":"","color":""}},{"type":"bold"},{"type":"italic"}],"text":"The Challenge of De Novo Binder Design"}]},{"type":"paragraph","attrs":{"id":"50f9011d-3d9c-4354-827c-194dde66cbec","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":"","color":""}}],"text":"Designing proteins that can selectively bind to molecular targets — like antibodies that block cancer pathways or neutralize pathogens — is one of the grand challenges in bioengineering. Traditional approaches are often slow, expensive, and dependent on animal immunization or brute-force screening of vast experimental libraries."}]},{"type":"paragraph","attrs":{"id":"ad8652c8-d625-4135-93a1-4df9f574ed51","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":"","color":""}}],"text":"But what if we could design these proteins directly from code?"}]},{"type":"paragraph","attrs":{"id":"708a3868-c374-4a2d-a9b1-2423a3a970c3","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":"","color":""}}],"text":"Recent advances in AI, particularly models like AlphaFold, RFdiffusion, and RFAntibody, have brought us closer to this vision. However, most existing tools still struggle to control the 3D fold of the resulting protein, limiting their ability to produce compact, stable, and therapeutically relevant scaffolds such as VHH nanobodies."}]},{"type":"paragraph","attrs":{"id":"2a58d4c3-f574-421e-8040-e8aa8523909e","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":"","color":""}}],"text":"This is particularly important when targeting proteins like EGFR (Epidermal Growth Factor Receptor) — a well-characterized cancer target implicated in multiple tumor types. Its clinical relevance makes EGFR an ideal benchmark for evaluating the next generation of AI-driven protein design tools."}]},{"type":"paragraph","attrs":{"id":"f9af6a7c-38bb-4b30-bfd2-936b6361080d","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":"","color":""}},{"type":"bold"},{"type":"italic"}],"text":"Enter FoldCraft"}]},{"type":"paragraph","attrs":{"id":"7aed97f3-0b46-4f98-a2f0-c42e957467a3","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":"","color":""}}],"text":"FoldCraft introduces a new approach called fold-conditioned hallucination, which combines AlphaFold2-Multimer’s generative capabilities with a single, interpretable contact map loss to guide structure formation. Rather than producing arbitrary shapes, FoldCraft steers sequence design toward antibody-like folds, enabling precise control over structural topology from the start."}]},{"type":"paragraph","attrs":{"id":"1f063c3d-b3db-432d-ad5c-5d92860a7018","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":"","color":""}}],"text":"In benchmarking studies, this simple yet powerful method leads to dramatic improvements. FoldCraft has outperformed state-of-the-art models like RFdiffusion and RFAntibody across multiple targets, achieving up to 19.5% success rates in generating valid binders against EGFR, PD-1, and PD-L1 — compared to less than 1% in many competing models."}]},{"type":"paragraph","attrs":{"id":"993bf57e-3f6d-4624-992e-6d7147442aa2","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":"","color":""}},{"type":"bold"},{"type":"italic"}],"text":"Why This Project Matters"}]},{"type":"paragraph","attrs":{"id":"6da1f3d1-8541-4bd0-a5aa-07719b7a64ff","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":"","color":""}}],"text":"We believe FoldCraft’s nanobody designs are ready for experimental testing. This project will determine:"}]},{"type":"paragraph","attrs":{"id":"50d80d9c-4f3e-48ce-8745-273408dc7d41","class":"MsoListParagraphCxSpFirst","textAlign":"justify"},"content":[{"type":"text","text":"·"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"7pt","fontFamily":"","color":""}}],"text":" "},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":"","color":""}}],"text":"Whether FoldCraft’s in silico success translates to real-world binding."}]},{"type":"paragraph","attrs":{"id":"9da31fda-e494-41ed-a7b3-414bc1f953f7","class":"MsoListParagraphCxSpMiddle","textAlign":"justify"},"content":[{"type":"text","text":"·"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"7pt","fontFamily":"","color":""}}],"text":" "},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":"","color":""}}],"text":"Whether AlphaFold3 confidence scores predict actual binding."}]},{"type":"paragraph","attrs":{"id":"c76369f6-4a07-483b-8413-3c698576bf23","class":"MsoListParagraphCxSpLast","textAlign":"justify"},"content":[{"type":"text","text":"·"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"7pt","fontFamily":"","color":""}}],"text":" "},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":"","color":""}}],"text":"Whether we can publicly release a dataset of fully de novo nanobody sequences and measured affinities — something rarely done in the field."}]},{"type":"paragraph","attrs":{"id":"501d9b20-b52a-45f6-8ba9-3034160509a1","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":"","color":""}},{"type":"bold"},{"type":"italic"}],"text":"Key Hypotheses"}]},{"type":"paragraph","attrs":{"id":"9d9ab480-6793-4a2b-acc1-dba8e97a9e1f","class":"MsoListParagraphCxSpFirst","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":"","color":""}}],"text":"H1: FoldCraft-designed nanobodies will show measurable binding to EGFR in wet-lab screening."}]},{"type":"paragraph","attrs":{"id":"7a2d2d37-7f65-408c-9bbf-d5684eb36c3a","class":"MsoListParagraphCxSpMiddle","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":"","color":""}}],"text":"H2: AlphaFold3 metrics (ipTM, pLDDT, iPAE) correlate with binding strength."}]},{"type":"paragraph","attrs":{"id":"9693f79f-a668-4739-9b6b-c340308750b4","class":"MsoListParagraphCxSpLast","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":"","color":""}}],"text":"H3: Optimization of only non-interface regions with SolMPNN will improve experimental binding compared to existing strategies."}]},{"type":"heading","attrs":{"id":"275805e1-4c5d-41c8-84e4-81496c76c950","data-toc-id":"275805e1-4c5d-41c8-84e4-81496c76c950","textAlign":"left","level":2},"content":[{"type":"text","text":"3. Methods"}]},{"type":"heading","attrs":{"id":"b1856b3e-8765-464b-811c-7006530ed0ef","data-toc-id":"b1856b3e-8765-464b-811c-7006530ed0ef","textAlign":"left","level":3},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}},{"type":"bold"},{"type":"italic"}],"text":"Sample"}]},{"type":"paragraph","attrs":{"id":"031c2453-affd-4b2e-bf7f-723253296c45","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"We will generate a large initial set of at least 500 nanobody sequences against EGFR using FoldCraft, focusing on structural designs that adhere to a canonical VHH fold. All sequences will undergo interface-preserving optimization using SolMPNN, and only those passing strict structural filters (e.g., AlphaFold3 plDDT > 70, ipAE < 10, ipTM > 0.5) will be retained."}]},{"type":"paragraph","attrs":{"id":"6cfd8dd5-bc3a-488d-859f-0059cde1a1d4","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"From this filtered pool, we will select the top 50 designs based on AlphaFold3 confidence metrics and sequence/epitope diversity for submission to experimental screening."}]},{"type":"paragraph","attrs":{"id":"29d34141-58da-4737-a57e-9a0fcd9cc368","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"In addition to high-throughput binding screens, we will "},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}},{"type":"bold"}],"text":"order the top-ranked (Top-1) FoldCraft design"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":" and perform a "},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}},{"type":"bold"}],"text":"functional validation in EGFR-positive cancer cell lines"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":" using standard cell-viability assays to assess antiproliferative effects. These experiments will be conducted with existing institutional resources and/or in-kind support without altering the $5,000 allocation to Adaptyv Bio."}]},{"type":"imageBlock","attrs":{"src":"https://storage.prod.researchhub.com/uploads/notes/users/180556/e37c4ccb-adb0-4c64-bad0-e6b723f992fd/fig1-1.png","width":"100%","align":"center"}},{"type":"paragraph","attrs":{"id":"4302150e-9e32-4bba-9abc-55174fdd8af5","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}},{"type":"bold"},{"type":"italic"}],"text":"Procedures"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"1. Step 1: In Silico Design"}]},{"type":"paragraph","attrs":{"id":"2f275a2d-b813-422b-99b1-82e81cf76f51","class":"MsoListParagraphCxSpFirst","textAlign":"justify"},"content":[{"type":"text","text":"·"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"7pt","fontFamily":null,"color":""}}],"text":" "},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"Generate nanobody scaffolds conditioned on the VHH fold."}]},{"type":"paragraph","attrs":{"id":"adc80019-b817-4d77-9d4f-ea1fac9a4ec1","class":"MsoListParagraphCxSpMiddle","textAlign":"justify"},"content":[{"type":"text","text":"·"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"7pt","fontFamily":null,"color":""}}],"text":" "},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"Define contact regions based on known EGFR epitopes."}]},{"type":"paragraph","attrs":{"id":"918b5f65-4cf8-412a-a276-38b585835186","class":"MsoListParagraphCxSpLast","textAlign":"justify"},"content":[{"type":"text","text":"·"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"7pt","fontFamily":null,"color":""}}],"text":" "},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"Optimize sequences using SolMPNN."}]},{"type":"paragraph","attrs":{"id":"3a391b3c-39b3-4ef0-8feb-7b428aba9bda","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"2. Step 2: Selection and Submission:"}]},{"type":"paragraph","attrs":{"id":"99851ea7-538a-4e85-8f32-65f07c1b537b","class":"MsoListParagraphCxSpFirst","textAlign":"justify"},"content":[{"type":"text","text":"·"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"7pt","fontFamily":null,"color":""}}],"text":" "},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"Rank designs by AlphaFold3 metrics."}]},{"type":"paragraph","attrs":{"id":"1fe18a82-292b-44cd-aabd-b84bf96208c4","class":"MsoListParagraphCxSpLast","textAlign":"justify"},"content":[{"type":"text","text":"·"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"7pt","fontFamily":null,"color":""}}],"text":" "},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"Cluster for diversity and select 50 designs."}]},{"type":"paragraph","attrs":{"id":"b695e9a5-ea63-4f14-b4a1-5c0a32c7d84e","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"3. Step 3: Experimental Screening (Adaptyv Bio):"}]},{"type":"paragraph","attrs":{"id":"4bbe5497-e226-480f-a238-86ecda2b9122","class":"MsoListParagraphCxSpFirst","textAlign":"justify"},"content":[{"type":"text","text":"·"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"7pt","fontFamily":null,"color":""}}],"text":" "},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"Submit all 50 sequences to Adaptyv Bio, who will synthesize, express, and test them against EGFR."}]},{"type":"paragraph","attrs":{"id":"0577c75c-fa64-45d6-b72e-bfb1a31ed4f7","class":"MsoListParagraphCxSpLast","textAlign":"justify"},"content":[{"type":"text","text":"·"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"7pt","fontFamily":null,"color":""}}],"text":" "},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"Results will include quantitative binding scores."}]},{"type":"paragraph","attrs":{"id":"b567fcb0-42c7-47a0-b247-62b0fc2f4695","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"4. Step 4: Data Analysis:"}]},{"type":"paragraph","attrs":{"id":"e7f421ed-4b5c-436b-bfb0-17acd340e5e1","class":"MsoListParagraphCxSpFirst","textAlign":"justify"},"content":[{"type":"text","text":"·"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"7pt","fontFamily":null,"color":""}}],"text":" "},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"Compute correlations between AlphaFold3 predictions and experimental scores."}]},{"type":"paragraph","attrs":{"id":"5a95cf55-9145-43b9-aff4-c952e5a0b3bc","class":"MsoListParagraphCxSpLast","textAlign":"justify"},"content":[{"type":"text","text":"·"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"7pt","fontFamily":null,"color":""}}],"text":" "},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"Visualize all structures, docking geometries, and experimental rankings."}]},{"type":"paragraph","attrs":{"id":"b17aaee2-1499-41b8-9677-149826e265dc","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}},{"type":"bold"}],"text":"Ethics and Data Management"}]},{"type":"paragraph","attrs":{"id":"63f731ba-42c1-4a77-995e-ffc6bc2a2e61","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"Cell-line work will follow institutional biosafety policies; authenticated lines and mycoplasma-free status will be documented. All raw viability data, analysis scripts, and the Top-1 sequence will be "},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}},{"type":"bold"}],"text":"open-sourced"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":" alongside binding data. All sequences, AlphaFold3 models, and binding data will be:"}]},{"type":"paragraph","attrs":{"id":"144dbb26-9c65-42a8-bdd0-5194c0032f45","class":"MsoListParagraphCxSpFirst","textAlign":"justify"},"content":[{"type":"text","text":"·"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"7pt","fontFamily":null,"color":""}}],"text":" "},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"Uploaded to GitHub (FoldCraft repo)"}]},{"type":"paragraph","attrs":{"id":"8a545aff-c72e-4f98-ac3a-d72b4d49b675","class":"MsoListParagraphCxSpMiddle","textAlign":"justify"},"content":[{"type":"text","text":"·"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"7pt","fontFamily":null,"color":""}}],"text":" "},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"Published on Zenodo with open CC-BY licensing"}]},{"type":"paragraph","attrs":{"id":"2a2f17e3-cdcd-4852-8e14-e6c4119207d0","class":"MsoListParagraphCxSpLast","textAlign":"justify"},"content":[{"type":"text","text":"·"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"7pt","fontFamily":null,"color":""}}],"text":" "},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"Included in a future preprint or publication"}]},{"type":"heading","attrs":{"id":"50236b91-422b-493d-aef1-c1e25b4f48c4","data-toc-id":"50236b91-422b-493d-aef1-c1e25b4f48c4","textAlign":"left","level":2},"content":[{"type":"text","text":"4. Pilot Data"}]},{"type":"paragraph","attrs":{"id":"f6254412-d5d0-4490-b914-1cc54c3090d8","class":null,"textAlign":"left"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"We have already validated FoldCraft computationally and published a preprint:"}]},{"type":"paragraph","attrs":{"id":"3b3fae7f-8763-4e01-a4dc-5c108dbc7311","class":"MsoListParagraphCxSpFirst","textAlign":"justify"},"content":[{"type":"text","text":"·"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"7pt","fontFamily":null,"color":""}}],"text":" "},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"Designed over 160 EGFR nanobody binders."}]},{"type":"paragraph","attrs":{"id":"8407bf4e-a76a-4cc9-81a5-746b5b141b62","class":"MsoListParagraphCxSpMiddle","textAlign":"justify"},"content":[{"type":"text","text":"·"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"7pt","fontFamily":null,"color":""}}],"text":" "},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"Achieved >19% success under AlphaFold3 evaluation thresholds."}]},{"type":"paragraph","attrs":{"id":"38409ec6-22d5-4e94-ac1c-363ecc0c9b08","class":"MsoListParagraphCxSpLast","textAlign":"justify"},"content":[{"type":"text","text":"·"},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"7pt","fontFamily":null,"color":""}}],"text":" "},{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"Outperformed RFAntibody (SOTA model), which failed to generate binders passing ipAE or ipTM cutoffs on diverse set of therapeutically relevant targets (PD-L1, PD1, EGFR, IFNAR)."}]},{"type":"imageBlock","attrs":{"src":"https://storage.prod.researchhub.com/uploads/notes/users/180556/a90c31ee-81d5-473c-aa21-387724163b74/fig4-1.png","width":"100%","align":"center"}},{"type":"paragraph","attrs":{"id":"b55337dd-0682-415b-8b77-96fdcd0962a4","class":"MsoListParagraphCxSpLast","textAlign":"justify"}},{"type":"paragraph","attrs":{"id":"ccf23fd1-2424-4666-a265-8021691af9f3","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"Additionally, our pipeline placed 2nd out of 74 submissions in the Adaptyv Bio EGFR binder design competition, by designing de novo EGFR miniproteins binder with KD of 4.77*10-6 M."}]},{"type":"paragraph","attrs":{"id":"037f1b20-47cb-42f2-98e8-44161e889af6","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"This proposal builds directly on that momentum."}]},{"type":"heading","attrs":{"id":"fefdd03b-f00d-4400-9bc7-f5fa34fb09ca","data-toc-id":"fefdd03b-f00d-4400-9bc7-f5fa34fb09ca","textAlign":"left","level":2},"content":[{"type":"text","text":"5. Budget"}]},{"type":"paragraph","attrs":{"id":"0d759746-fb7d-4c0e-8f2f-16ccdc832a41","class":null,"textAlign":"left"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"All funds will be used to run 50 binding screens at Adaptyv Bio."}]},{"type":"table","attrs":{"id":"4b7f3263-0950-42d2-a36d-d68526a08941"},"content":[{"type":"tableRow","content":[{"type":"tableCell","attrs":{"colspan":1,"rowspan":1,"colwidth":null,"style":"width:2.0in;border:solid windowtext 1.0pt;\n mso-border-alt:solid windowtext .5pt;padding:0in 5.4pt 0in 5.4pt"},"content":[{"type":"paragraph","attrs":{"id":"14feabc7-5b65-4a97-82f7-562f15f2ba53","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"bold"}],"text":"Item"}]}]},{"type":"tableCell","attrs":{"colspan":1,"rowspan":1,"colwidth":null,"style":"width:2.0in;border:solid windowtext 1.0pt;\n border-left:none;mso-border-left-alt:solid windowtext .5pt;mso-border-alt:\n solid windowtext .5pt;padding:0in 5.4pt 0in 5.4pt"},"content":[{"type":"paragraph","attrs":{"id":"75267a8f-ca16-444f-9441-a89cdd71818e","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"bold"}],"text":"Description"}]}]},{"type":"tableCell","attrs":{"colspan":1,"rowspan":1,"colwidth":null,"style":"width:2.0in;border:solid windowtext 1.0pt;\n border-left:none;mso-border-left-alt:solid windowtext .5pt;mso-border-alt:\n solid windowtext .5pt;padding:0in 5.4pt 0in 5.4pt"},"content":[{"type":"paragraph","attrs":{"id":"97c61cf1-9e57-4741-a1fa-73f9747ea1b0","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"bold"}],"text":"Cost (USD)"}]}]}]},{"type":"tableRow","content":[{"type":"tableCell","attrs":{"colspan":1,"rowspan":1,"colwidth":null,"style":"width:2.0in;border:solid windowtext 1.0pt;\n border-top:none;mso-border-top-alt:solid windowtext .5pt;mso-border-alt:solid windowtext .5pt;\n padding:0in 5.4pt 0in 5.4pt"},"content":[{"type":"paragraph","attrs":{"id":"b0e73e55-bdd2-457c-91c0-fc227b7b35a0","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","text":"Adaptyv Bio binder screening"}]}]},{"type":"tableCell","attrs":{"colspan":1,"rowspan":1,"colwidth":null,"style":"width:2.0in;border-top:none;border-left:none;\n border-bottom:solid windowtext 1.0pt;border-right:solid windowtext 1.0pt;\n mso-border-top-alt:solid windowtext .5pt;mso-border-left-alt:solid windowtext .5pt;\n mso-border-alt:solid windowtext .5pt;padding:0in 5.4pt 0in 5.4pt"},"content":[{"type":"paragraph","attrs":{"id":"5c0b13fc-7325-47b3-b703-7c9c4a7e34f9","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","text":"50 de novo nanobody sequences vs. EGFR"}]}]},{"type":"tableCell","attrs":{"colspan":1,"rowspan":1,"colwidth":null,"style":"width:2.0in;border-top:none;border-left:none;\n border-bottom:solid windowtext 1.0pt;border-right:solid windowtext 1.0pt;\n mso-border-top-alt:solid windowtext .5pt;mso-border-left-alt:solid windowtext .5pt;\n mso-border-alt:solid windowtext .5pt;padding:0in 5.4pt 0in 5.4pt"},"content":[{"type":"paragraph","attrs":{"id":"7ddefae1-e326-4e0b-aaa5-14c7a0b8e28e","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","text":"$5000"}]}]}]},{"type":"tableRow","content":[{"type":"tableCell","attrs":{"colspan":1,"rowspan":1,"colwidth":null,"style":"width:2.0in;border:solid windowtext 1.0pt;\n border-top:none;mso-border-top-alt:solid windowtext .5pt;mso-border-alt:solid windowtext .5pt;\n padding:0in 5.4pt 0in 5.4pt"},"content":[{"type":"paragraph","attrs":{"id":"6776658e-9b1c-4513-b430-f6111d8fecde","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"bold"}],"text":"Total"}]}]},{"type":"tableCell","attrs":{"colspan":1,"rowspan":1,"colwidth":null,"style":"width:2.0in;border-top:none;border-left:none;\n border-bottom:solid windowtext 1.0pt;border-right:solid windowtext 1.0pt;\n mso-border-top-alt:solid windowtext .5pt;mso-border-left-alt:solid windowtext .5pt;\n mso-border-alt:solid windowtext .5pt;padding:0in 5.4pt 0in 5.4pt"},"content":[{"type":"paragraph","attrs":{"id":"13c2853c-0ae8-4a7b-8891-8f96d4fa2f98","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"bold"}],"text":" "}]}]},{"type":"tableCell","attrs":{"colspan":1,"rowspan":1,"colwidth":null,"style":"width:2.0in;border-top:none;border-left:none;\n border-bottom:solid windowtext 1.0pt;border-right:solid windowtext 1.0pt;\n mso-border-top-alt:solid windowtext .5pt;mso-border-left-alt:solid windowtext .5pt;\n mso-border-alt:solid windowtext .5pt;padding:0in 5.4pt 0in 5.4pt"},"content":[{"type":"paragraph","attrs":{"id":"4ca8f240-6de8-42cf-aadd-7d867588369c","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"bold"}],"text":"$5000"}]}]}]}]},{"type":"paragraph","attrs":{"id":"33e7af2f-31a6-4b0b-a631-03c0a1eea96e","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"Computational modeling is performed on existing local GPU infrastructure (RTX 3090/4080/4090)."}]},{"type":"heading","attrs":{"id":"771a83fc-c1ab-45d8-bf50-9e87602cd9cf","data-toc-id":"771a83fc-c1ab-45d8-bf50-9e87602cd9cf","textAlign":"left","level":2},"content":[{"type":"text","text":"6. References"}]},{"type":"paragraph","attrs":{"id":"ed1998c9-6268-4ff7-89dd-b7bf9dcb2259","class":null,"textAlign":"left"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"1. Rustamov, K.R., and Baev, A.Y. Fold-Conditioned De Novo Binder Design via AlphaFold2-Multimer Hallucination. bioRxiv, 2025."}]},{"type":"paragraph","attrs":{"id":"ac7518dd-04d1-4a51-abf1-875fb762fd5e","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"2. Cotet, T.-S., et al. Crowdsourced Protein Design: Lessons From the Adaptyv EGFR Binder Competition. bioRxiv, 2025."}]},{"type":"paragraph","attrs":{"id":"a2a04e1a-7936-4180-a737-381dd3fdf1ca","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"3. Jumper, J. et al. Highly accurate protein structure prediction with AlphaFold. Nature, 2021."}]},{"type":"paragraph","attrs":{"id":"2741a462-92e7-4b0f-806c-4f48d1880cbd","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"4. Dauparas, J. et al. Robust deep learning-based protein sequence design using ProteinMPNN. Science, 2022."}]},{"type":"paragraph","attrs":{"id":"bab49267-2936-4518-a055-791cb98f634d","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"5. Bennett, N.R. et al. Atomically accurate de novo design of antibodies with RFdiffusion. bioRxiv, 2025."}]},{"type":"paragraph","attrs":{"id":"bf1f8157-9c12-4b8e-9eb2-d5d31a6d026c","class":"MsoNormal","textAlign":"justify"},"content":[{"type":"text","marks":[{"type":"textStyle","attrs":{"fontSize":"12pt","fontFamily":null,"color":""}}],"text":"6. Abramson, J. et al. Accurate structure prediction of biomolecular interactions with AlphaFold3. Nature, 2024."}]}]}5b:T5665,

FoldCraft: An Open Framework for De Novo Design of EGFR-Targeting Nanobodies

1. Overview

This project aims to experimentally validate nanobody binders against EGFR, designed entirely from scratch using our AI-powered platform, FoldCraft. FoldCraft uses AlphaFold2-Multimer as a generative engine and has already demonstrated performance superior to state-of-the-art (SOTA) models like RFdiffusion and RFAntibody in in silico benchmarks.

To test these designs in real-world settings, we will submit our top 50 nanobody candidates to the Adaptyv Bio binding screening platform, a high-throughput experimental assay that rapidly evaluates the binding ability of synthetic proteins.

This work builds on a strong foundation: FoldCraft placed 2nd in Round 1 of Adaptyv Bio’s EGFR binder design competition, showing its potential to generate high-quality miniproteins binders with therapeutic relevance.

The project will be completed in 6 months, with no additional infrastructure or ethics approvals required. Most importantly, we commit to open-sourcing all binder sequences and binding data — making this a fully transparent and reusable resource for the broader scientific community.

2. Introduction

The Challenge of De Novo Binder Design

Designing proteins that can selectively bind to molecular targets — like antibodies that block cancer pathways or neutralize pathogens — is one of the grand challenges in bioengineering. Traditional approaches are often slow, expensive, and dependent on animal immunization or brute-force screening of vast experimental libraries.

But what if we could design these proteins directly from code?

Recent advances in AI, particularly models like AlphaFold, RFdiffusion, and RFAntibody, have brought us closer to this vision. However, most existing tools still struggle to control the 3D fold of the resulting protein, limiting their ability to produce compact, stable, and therapeutically relevant scaffolds such as VHH nanobodies.

This is particularly important when targeting proteins like EGFR (Epidermal Growth Factor Receptor) — a well-characterized cancer target implicated in multiple tumor types. Its clinical relevance makes EGFR an ideal benchmark for evaluating the next generation of AI-driven protein design tools.

Enter FoldCraft

FoldCraft introduces a new approach called fold-conditioned hallucination, which combines AlphaFold2-Multimer’s generative capabilities with a single, interpretable contact map loss to guide structure formation. Rather than producing arbitrary shapes, FoldCraft steers sequence design toward antibody-like folds, enabling precise control over structural topology from the start.

In benchmarking studies, this simple yet powerful method leads to dramatic improvements. FoldCraft has outperformed state-of-the-art models like RFdiffusion and RFAntibody across multiple targets, achieving up to 19.5% success rates in generating valid binders against EGFR, PD-1, and PD-L1 — compared to less than 1% in many competing models.

Why This Project Matters

We believe FoldCraft’s nanobody designs are ready for experimental testing. This project will determine:

· Whether FoldCraft’s in silico success translates to real-world binding.

· Whether AlphaFold3 confidence scores predict actual binding.

· Whether we can publicly release a dataset of fully de novo nanobody sequences and measured affinities — something rarely done in the field.

Key Hypotheses

H1: FoldCraft-designed nanobodies will show measurable binding to EGFR in wet-lab screening.

H2: AlphaFold3 metrics (ipTM, pLDDT, iPAE) correlate with binding strength.

H3: Optimization of only non-interface regions with SolMPNN will improve experimental binding compared to existing strategies.

3. Methods

Sample

We will generate a large initial set of at least 500 nanobody sequences against EGFR using FoldCraft, focusing on structural designs that adhere to a canonical VHH fold. All sequences will undergo interface-preserving optimization using SolMPNN, and only those passing strict structural filters (e.g., AlphaFold3 plDDT > 70, ipAE < 10, ipTM > 0.5) will be retained.

From this filtered pool, we will select the top 50 designs based on AlphaFold3 confidence metrics and sequence/epitope diversity for submission to experimental screening.

In addition to high-throughput binding screens, we will order the top-ranked (Top-1) FoldCraft design and perform a functional validation in EGFR-positive cancer cell lines using standard cell-viability assays to assess antiproliferative effects. These experiments will be conducted with existing institutional resources and/or in-kind support without altering the $5,000 allocation to Adaptyv Bio.

Procedures1. Step 1: In Silico Design

· Generate nanobody scaffolds conditioned on the VHH fold.

· Define contact regions based on known EGFR epitopes.

· Optimize sequences using SolMPNN.

2. Step 2: Selection and Submission:

· Rank designs by AlphaFold3 metrics.

· Cluster for diversity and select 50 designs.

3. Step 3: Experimental Screening (Adaptyv Bio):

· Submit all 50 sequences to Adaptyv Bio, who will synthesize, express, and test them against EGFR.

· Results will include quantitative binding scores.

4. Step 4: Data Analysis:

· Compute correlations between AlphaFold3 predictions and experimental scores.

· Visualize all structures, docking geometries, and experimental rankings.

Ethics and Data Management

Cell-line work will follow institutional biosafety policies; authenticated lines and mycoplasma-free status will be documented. All raw viability data, analysis scripts, and the Top-1 sequence will be open-sourced alongside binding data. All sequences, AlphaFold3 models, and binding data will be:

· Uploaded to GitHub (FoldCraft repo)

· Published on Zenodo with open CC-BY licensing

· Included in a future preprint or publication

4. Pilot Data

We have already validated FoldCraft computationally and published a preprint:

· Designed over 160 EGFR nanobody binders.

· Achieved >19% success under AlphaFold3 evaluation thresholds.

· Outperformed RFAntibody (SOTA model), which failed to generate binders passing ipAE or ipTM cutoffs on diverse set of therapeutically relevant targets (PD-L1, PD1, EGFR, IFNAR).

Additionally, our pipeline placed 2nd out of 74 submissions in the Adaptyv Bio EGFR binder design competition, by designing de novo EGFR miniproteins binder with KD of 4.77*10-6 M.

This proposal builds directly on that momentum.

5. Budget

All funds will be used to run 50 binding screens at Adaptyv Bio.

Item	Description	Cost (USD)
Adaptyv Bio binder screening	50 de novo nanobody sequences vs. EGFR	$5000
Total		$5000

Computational modeling is performed on existing local GPU infrastructure (RTX 3090/4080/4090).

6. References

1. Rustamov, K.R., and Baev, A.Y. Fold-Conditioned De Novo Binder Design via AlphaFold2-Multimer Hallucination. bioRxiv, 2025.

2. Cotet, T.-S., et al. Crowdsourced Protein Design: Lessons From the Adaptyv EGFR Binder Competition. bioRxiv, 2025.

3. Jumper, J. et al. Highly accurate protein structure prediction with AlphaFold. Nature, 2021.

4. Dauparas, J. et al. Robust deep learning-based protein sequence design using ProteinMPNN. Science, 2022.

5. Bennett, N.R. et al. Atomically accurate de novo design of antibodies with RFdiffusion. bioRxiv, 2025.

6. Abramson, J. et al. Accurate structure prediction of biomolecular interactions with AlphaFold3. Nature, 2024.

5c:T5665,

FoldCraft: An Open Framework for De Novo Design of EGFR-Targeting Nanobodies

1. Overview

2. Introduction

The Challenge of De Novo Binder Design

But what if we could design these proteins directly from code?

Enter FoldCraft

Why This Project Matters

We believe FoldCraft’s nanobody designs are ready for experimental testing. This project will determine:

· Whether FoldCraft’s in silico success translates to real-world binding.

· Whether AlphaFold3 confidence scores predict actual binding.

· Whether we can publicly release a dataset of fully de novo nanobody sequences and measured affinities — something rarely done in the field.

Key Hypotheses

H1: FoldCraft-designed nanobodies will show measurable binding to EGFR in wet-lab screening.

H2: AlphaFold3 metrics (ipTM, pLDDT, iPAE) correlate with binding strength.

H3: Optimization of only non-interface regions with SolMPNN will improve experimental binding compared to existing strategies.

3. Methods

Sample

From this filtered pool, we will select the top 50 designs based on AlphaFold3 confidence metrics and sequence/epitope diversity for submission to experimental screening.

Procedures1. Step 1: In Silico Design

· Generate nanobody scaffolds conditioned on the VHH fold.

· Define contact regions based on known EGFR epitopes.

· Optimize sequences using SolMPNN.

2. Step 2: Selection and Submission:

· Rank designs by AlphaFold3 metrics.

· Cluster for diversity and select 50 designs.

3. Step 3: Experimental Screening (Adaptyv Bio):

· Submit all 50 sequences to Adaptyv Bio, who will synthesize, express, and test them against EGFR.

· Results will include quantitative binding scores.

4. Step 4: Data Analysis:

· Compute correlations between AlphaFold3 predictions and experimental scores.

· Visualize all structures, docking geometries, and experimental rankings.

Ethics and Data Management

· Uploaded to GitHub (FoldCraft repo)

· Published on Zenodo with open CC-BY licensing

· Included in a future preprint or publication

4. Pilot Data

We have already validated FoldCraft computationally and published a preprint:

· Designed over 160 EGFR nanobody binders.

· Achieved >19% success under AlphaFold3 evaluation thresholds.

· Outperformed RFAntibody (SOTA model), which failed to generate binders passing ipAE or ipTM cutoffs on diverse set of therapeutically relevant targets (PD-L1, PD1, EGFR, IFNAR).

Additionally, our pipeline placed 2nd out of 74 submissions in the Adaptyv Bio EGFR binder design competition, by designing de novo EGFR miniproteins binder with KD of 4.77*10-6 M.

This proposal builds directly on that momentum.

5. Budget

All funds will be used to run 50 binding screens at Adaptyv Bio.

Item	Description	Cost (USD)
Adaptyv Bio binder screening	50 de novo nanobody sequences vs. EGFR	$5000
Total		$5000

Computational modeling is performed on existing local GPU infrastructure (RTX 3090/4080/4090).

6. References

1. Rustamov, K.R., and Baev, A.Y. Fold-Conditioned De Novo Binder Design via AlphaFold2-Multimer Hallucination. bioRxiv, 2025.

2. Cotet, T.-S., et al. Crowdsourced Protein Design: Lessons From the Adaptyv EGFR Binder Competition. bioRxiv, 2025.

3. Jumper, J. et al. Highly accurate protein structure prediction with AlphaFold. Nature, 2021.

4. Dauparas, J. et al. Robust deep learning-based protein sequence design using ProteinMPNN. Science, 2022.

5. Bennett, N.R. et al. Atomically accurate de novo design of antibodies with RFdiffusion. bioRxiv, 2025.

6. Abramson, J. et al. Accurate structure prediction of biomolecular interactions with AlphaFold3. Nature, 2024.

5d:T40b,FoldCraft is a computational protein-design platform that uses AlphaFold2-Multimer with fold-conditioned hallucination to generate de novo EGFR-targeting nanobodies. The proposal seeks $5,000 to experimentally validate 50 top-ranked designs via Adaptyv Bio's high-throughput binding assay. The work builds on strong in silico benchmarking (19.5% success rate, 2nd place in Adaptyv EGFR competition) and commits to open-sourcing all sequences and binding data. Strengths include clear scientific rationale, alignment with current protein-design trends, and a well-defined experimental plan with realistic scope. Weaknesses include limited detail on statistical analysis, modest team track record (h-index 3-4), unclear institutional affiliation and wet-lab capacity, and lack of power analysis or multiple-comparison correction strategy. The proposal is well-motivated and feasible in scope but lacks rigor in statistical planning and does not clearly establish the team's capacity to execute functional validation in cancer cell lines.5e:T1e42,FoldCraft: An Open Framework for De Novo Design of EGFR-Targeting Nanobodies 1. Overview This project aims to experimentally validate nanobody binders against EGFR, designed entirely from scratch using our AI-powered platform, FoldCraft. FoldCraft uses AlphaFold2-Multimer as a generative engine and has already demonstrated performance superior to state-of-the-art (SOTA) models like RFdiffusion and RFAntibody in in silico benchmarks. To test these designs in real-world settings, we will submit our top 50 nanobody candidates to the Adaptyv Bio binding screening platform, a high-throughput experimental assay that rapidly evaluates the binding ability of synthetic proteins. This work builds on a strong foundation: FoldCraft placed 2nd in Round 1 of Adaptyv Bio’s EGFR binder design competition, showing its potential to generate high-quality miniproteins binders with therapeutic relevance. The project will be completed in 6 months, with no additional infrastructure or ethics approvals required. Most importantly, we commit to open-sourcing all binder sequences and binding data — making this a fully transparent and reusable resource for the broader scientific community. 2. Introduction The Challenge of De Novo Binder Design Designing proteins that can selectively bind to molecular targets — like antibodies that block cancer pathways or neutralize pathogens — is one of the grand challenges in bioengineering. Traditional approaches are often slow, expensive, and dependent on animal immunization or brute-force screening of vast experimental libraries. But what if we could design these proteins directly from code? Recent advances in AI, particularly models like AlphaFold, RFdiffusion, and RFAntibody, have brought us closer to this vision. However, most existing tools still struggle to control the 3D fold of the resulting protein, limiting their ability to produce compact, stable, and therapeutically relevant scaffolds such as VHH nanobodies. This is particularly important when targeting proteins like EGFR (Epidermal Growth Factor Receptor) — a well-characterized cancer target implicated in multiple tumor types. Its clinical relevance makes EGFR an ideal benchmark for evaluating the next generation of AI-driven protein design tools. Enter FoldCraft FoldCraft introduces a new approach called fold-conditioned hallucination, which combines AlphaFold2-Multimer’s generative capabilities with a single, interpretable contact map loss to guide structure formation. Rather than producing arbitrary shapes, FoldCraft steers sequence design toward antibody-like folds, enabling precise control over structural topology from the start. In benchmarking studies, this simple yet powerful method leads to dramatic improvements. FoldCraft has outperformed state-of-the-art models like RFdiffusion and RFAntibody across multiple targets, achieving up to 19.5% success rates in generating valid binders against EGFR, PD-1, and PD-L1 — compared to less than 1% in many competing models. Why This Project Matters We believe FoldCraft’s nanobody designs are ready for experimental testing. This project will determine: · Whether FoldCraft’s in silico success translates to real-world binding. · Whether AlphaFold3 confidence scores predict actual binding. · Whether we can publicly release a dataset of fully de novo nanobody sequences and measured affinities — something rarely done in the field. Key Hypotheses H1: FoldCraft-designed nanobodies will show measurable binding to EGFR in wet-lab screening. H2: AlphaFold3 metrics (ipTM, pLDDT, iPAE) correlate with binding strength. H3: Optimization of only non-interface regions with SolMPNN will improve experimental binding compared to existing strategies. 3. Methods Sample We will generate a large initial set of at least 500 nanobody sequences against EGFR using FoldCraft, focusing on structural designs that adhere to a canonical VHH fold. All sequences will undergo interface-preserving optimization using SolMPNN, and only those passing strict structural filters (e.g., AlphaFold3 plDDT > 70, ipAE < 10, ipTM > 0.5) will be retained. From this filtered pool, we will select the top 50 designs based on AlphaFold3 confidence metrics and sequence/epitope diversity for submission to experimental screening. In addition to high-throughput binding screens, we will order the top-ranked (Top-1) FoldCraft design and perform a functional validation in EGFR-positive cancer cell lines using standard cell-viability assays to assess antiproliferative effects. These experiments will be conducted with existing institutional resources and/or in-kind support without altering the $5,000 allocation to Adaptyv Bio. Procedures1. Step 1: In Silico Design · Generate nanobody scaffolds conditioned on the VHH fold. · Define contact regions based on known EGFR epitopes. · Optimize sequences using SolMPNN. 2. Step 2: Selection and Submission: · Rank designs by AlphaFold3 metrics. · Cluster for diversity and select 50 designs. 3. Step 3: Experimental Screening (Adaptyv Bio): · Submit all 50 sequences to Adaptyv Bio, who will synthesize, express, and test them against EGFR. · Results will include quantitative binding scores. 4. Step 4: Data Analysis: · Compute correlations between AlphaFold3 predictions and experimental scores. · Visualize all structures, docking geometries, and experimental rankings. Ethics and Data Management Cell-line work will follow institutional biosafety policies; authenticated lines and mycoplasma-free status will be documented. All raw viability data, analysis scripts, and the Top-1 sequence will be open-sourced alongside binding data. All sequences, AlphaFold3 models, and binding data will be: · Uploaded to GitHub (FoldCraft repo) · Published on Zenodo with open CC-BY licensing · Included in a future preprint or publication 4. Pilot Data We have already validated FoldCraft computationally and published a preprint: · Designed over 160 EGFR nanobody binders. · Achieved >19% success under AlphaFold3 evaluation thresholds. · Outperformed RFAntibody (SOTA model), which failed to generate binders passing ipAE or ipTM cutoffs on diverse set of therapeutically relevant targets (PD-L1, PD1, EGFR, IFNAR). Additionally, our pipeline placed 2nd out of 74 submissions in the Adaptyv Bio EGFR binder design competition, by designing de novo EGFR miniproteins binder with KD of 4.77*10-6 M. This proposal builds directly on that momentum. 5. Budget All funds will be used to run 50 binding screens at Adaptyv Bio. Item Description Cost (USD) Adaptyv Bio binder screening 50 de novo nanobody sequences vs. EGFR $5000 Total $5000 Computational modeling is performed on existing local GPU infrastructure (RTX 3090/4080/4090). 6. References 1. Rustamov, K.R., and Baev, A.Y. Fold-Conditioned De Novo Binder Design via AlphaFold2-Multimer Hallucination. bioRxiv, 2025. 2. Cotet, T.-S., et al. Crowdsourced Protein Design: Lessons From the Adaptyv EGFR Binder Competition. bioRxiv, 2025. 3. Jumper, J. et al. Highly accurate protein structure prediction with AlphaFold. Nature, 2021. 4. Dauparas, J. et al. Robust deep learning-based protein sequence design using ProteinMPNN. Science, 2022. 5. Bennett, N.R. et al. Atomically accurate de novo design of antibodies with RFdiffusion. bioRxiv, 2025. 6. Abramson, J. et al. Accurate structure prediction of biomolecular interactions with AlphaFold3. Nature, 2024.21:["$","$L54",null,{"children":["$","$L4",null,{"topBanner":["$","$L55",null,{"work":{"id":4324,"type":"$undefined","contentType":"preregistration","title":"FoldCraft: An Open Framework for De Novo Design of EGFR-Targeting Nanobodies","slug":"foldcraft-an-open-framework-for-de-novo-design-of-egfr-targeting-nanobodies","createdDate":"2025-08-15T11:28:27.721283Z","createdByUserId":180556,"updatedDate":"2025-08-19T13:01:55.299933Z","publishedDate":"2025-08-15T11:28:27.721283Z","authors":[{"authorProfile":{"id":7944791,"fullName":"Khondamir Rustamov","firstName":"Khondamir","lastName":"Rustamov","profileImage":"https://lh3.googleusercontent.com/a/ACg8ocI9wKqhqtic0gXOB7ULMtEvdp4_maQzczSa2lLNtFiFc-KH_4TRVA=s96-c","headline":"","profileUrl":"/author/7944791","user":{"id":0,"email":"","firstName":"","lastName":"","fullName":"Unknown User","isVerified":false,"authorProfile":"$undefined","balance":0,"lockedBalance":0,"promotionalBalance":0,"fundingCredits":0,"balances":"$undefined","totalRsc":0,"hasCompletedOnboarding":false,"createdDate":"$undefined","moderator":false,"editorOfHubs":"$undefined","isModerator":false,"isFunder":false,"referralCode":"$undefined","authProvider":"$undefined","isStakingOptedIn":false,"balanceHistory":0,"raw":180556},"description":"$undefined","createdDate":"2025-08-04T08:33:59.841980Z","education":[],"twitter":"$undefined","facebook":"$undefined","linkedin":"https://www.linkedin.com/in/khondamir-rustamov-840118180/","googleScholar":"https://scholar.google.com/citations?user=GofC7xEAAAAJ&hl=en","orcidId":"https://orcid.org/0000-0001-7661-869X","isOrcidConnected":true,"isClaimed":true,"hIndex":3,"i10Index":"$undefined","userId":"$undefined","isVerified":true,"isHubEditor":false,"editorOfHubs":[],"raw":{"id":7944791,"is_public":true,"is_removed":false,"is_removed_date":null,"user":180556,"first_name":"Khondamir","last_name":"Rustamov","created_date":"2025-08-04T08:33:59.841980Z","updated_date":"2026-04-27T05:56:26.357541Z","description":"","h_index":3,"i10_index":0,"profile_image":"https://lh3.googleusercontent.com/a/ACg8ocI9wKqhqtic0gXOB7ULMtEvdp4_maQzczSa2lLNtFiFc-KH_4TRVA=s96-c","author_score":0,"university":null,"orcid_id":"https://orcid.org/0000-0001-7661-869X","openalex_ids":[],"education":"$21:props:children:props:topBanner:props:work:authors:0:authorProfile:education","headline":"","facebook":null,"twitter":"","linkedin":"https://www.linkedin.com/in/khondamir-rustamov-840118180/","google_scholar":"https://scholar.google.com/citations?user=GofC7xEAAAAJ&hl=en","claimed":true,"country_code":null,"created_source":"RESEARCHHUB","last_full_fetch_from_openalex":null,"two_year_mean_citedness":2.4,"merged_with_author":null,"added_as_editor_date":null,"is_hub_editor_of":[],"is_hub_editor":false,"num_posts":2,"is_orcid_connected":true,"reputation":100,"reputation_v2":null,"reputation_list":[],"suspended_status":{"probable_spammer":false,"is_suspended":false},"total_score":null,"wallet":null,"is_verified":true}},"isCorresponding":false,"position":"middle","raw":"$21:props:children:props:topBanner:props:work:authors:0:authorProfile:raw"}],"abstract":"$56","doi":"10.55277/researchhub.lo09s2v9","journal":"$undefined","topics":[{"id":7,"name":"Bioinformatics","slug":"bioinformatics"},{"id":455,"name":"Protein Design","slug":"protein-design"},{"id":61768,"name":"Cancer","slug":"cancer-1"}],"formats":[],"license":"$undefined","pdfCopyrightAllowsDisplay":true,"figures":[],"versions":[],"metrics":{"votes":6,"adjustedScore":"$undefined","comments":1,"saves":0,"reviewScore":0,"reviews":0,"earned":0,"views":0},"unifiedDocumentId":7555962,"registeredReportId":null,"postType":"PREREGISTRATION","fundraise":"$undefined","note":{"id":18252,"access":"WORKSPACE","organization":{"id":176241,"memberCount":1,"userPermission":"$undefined","createdDate":"2025-08-04T08:33:59.915641Z","coverImage":null,"description":"","name":"Khondamir Rustamov's Org","slug":"khondamir-rustamovs-org"},"createdDate":"2025-08-15T11:19:07.561135Z","updatedDate":"2025-08-15T11:24:22.303822Z","title":"FoldCraft: An Open Framework for De Novo Design of EGFR-Targeting Nanobodies","isRemoved":false,"post":{"id":4324,"slug":"foldcraft-an-open-framework-for-de-novo-design-of-egfr-targeting-nanobodies","contentType":"preregistration","documentType":"PREREGISTRATION","moderationStatus":"APPROVED","fundraise":{"id":105,"amountRaised":{"usd":0.64648207,"rsc":10},"goalAmount":{"usd":5000,"rsc":77341.66548501492},"contributors":{"numContributors":1,"topContributors":[{"id":31620,"authorProfile":{"id":928488,"fullName":"Ioana Gheorghe","firstName":"Ioana","lastName":"Gheorghe","profileImage":"https://storage.prod.researchhub.com/uploads/author_profile_images/2022/02/02/blob_xuIQVGV","headline":"","profileUrl":"/author/928488","user":{"id":0,"email":"","firstName":"","lastName":"","fullName":"Unknown User","isVerified":false,"authorProfile":"$undefined","balance":0,"lockedBalance":0,"promotionalBalance":0,"fundingCredits":0,"balances":"$undefined","totalRsc":0,"hasCompletedOnboarding":false,"createdDate":"$undefined","moderator":false,"editorOfHubs":"$undefined","isModerator":false,"isFunder":false,"referralCode":"$undefined","authProvider":"$undefined","isStakingOptedIn":false,"balanceHistory":0,"raw":31620},"description":"$undefined","createdDate":"2021-07-13T01:05:04.258003Z","education":[],"twitter":"$undefined","facebook":"$undefined","linkedin":"$undefined","googleScholar":"$undefined","orcidId":"$undefined","isOrcidConnected":false,"isClaimed":true,"hIndex":"$undefined","i10Index":"$undefined","userId":"$undefined","isVerified":true,"isHubEditor":false,"editorOfHubs":[],"raw":{"id":928488,"is_verified":true,"is_public":true,"is_removed":false,"is_removed_date":null,"first_name":"Ioana","last_name":"Gheorghe","created_date":"2021-07-13T01:05:04.258003Z","updated_date":"2022-06-09T10:13:41.269811Z","description":"","h_index":0,"i10_index":0,"profile_image":"https://storage.prod.researchhub.com/uploads/author_profile_images/2022/02/02/blob_xuIQVGV","author_score":0,"orcid_id":null,"openalex_ids":[],"education":"$21:props:children:props:topBanner:props:work:note:post:fundraise:contributors:topContributors:0:authorProfile:education","headline":"","facebook":null,"twitter":null,"linkedin":null,"google_scholar":null,"claimed":true,"country_code":null,"created_source":"RESEARCHHUB","last_full_fetch_from_openalex":null,"two_year_mean_citedness":0,"user":31620,"university":null,"merged_with_author":null}},"totalContribution":{"rsc":10,"usd":0},"contributions":[{"amount":10,"currency":"RSC","date":"2025-08-18T16:02:59.574195Z"}]}]},"createdBy":{"id":0,"email":"","firstName":"","lastName":"","fullName":"Unknown User","isVerified":false,"authorProfile":"$undefined","balance":0,"lockedBalance":0,"totalRsc":0,"hasCompletedOnboarding":false,"moderator":false,"raw":null,"isModerator":false,"isFunder":false,"authProvider":"$undefined","isStakingOptedIn":false,"balanceHistory":0},"status":"OPEN","goalCurrency":"USD","startDate":"2025-08-15T11:28:27.794949Z","endDate":"2025-09-14T11:28:27.794822Z","createdDate":"","updatedDate":"","postId":"$undefined","postTitle":"$undefined","postSlug":"$undefined","postImage":null,"reviewMetrics":"$undefined","nonprofit":"$undefined","raw":{"id":105,"amount_raised":{"usd":0.64648207,"rsc":10},"goal_amount":{"usd":5000,"rsc":77341.66548501492},"contributors":{"total":1,"top":[{"id":31620,"author_profile":"$21:props:children:props:topBanner:props:work:note:post:fundraise:contributors:topContributors:0:authorProfile:raw","is_funder":false,"rsc_earned":null,"balances":null,"benefits_expire_on":null,"editor_of":[],"is_verified":true,"last_login":"2026-07-30T23:57:22.187546Z","is_superuser":false,"username":"ioanagheorghe.x@gmail.com","first_name":"Ioana","last_name":"Gheorghe","email":"ioanagheorghe.x@gmail.com","is_staff":false,"is_active":true,"date_joined":"2021-07-13T01:05:04Z","is_public":true,"is_removed":false,"is_removed_date":null,"agreed_to_terms":false,"clicked_on_balance_date":"2023-02-22T22:44:03.177438Z","country_code":"RO","created_date":"2021-07-13T01:05:04.176891Z","has_seen_first_coin_modal":true,"has_seen_orcid_connect_modal":true,"has_completed_onboarding":true,"is_suspended":false,"moderator":false,"probable_spammer":false,"referral_code":"82e42395-35d8-4770-a182-2d01554a50eb","reputation":268,"should_display_rsc_balance_home":true,"is_staking_opted_in":true,"staking_opted_in_date":"2026-04-16T15:44:08.328272Z","spam_updated_date":null,"suspended_updated_date":"2023-02-24T07:36:58.317667Z","updated_date":"2025-08-18T13:01:47.353174Z","upload_tutorial_complete":false,"is_official_account":false,"invited_by":34416,"groups":[],"user_permissions":[],"total_contribution":{"rsc":10,"rsc_usd_snapshot":5.664491989,"usd":0},"contributions":[{"amount":10,"currency":"RSC","date":"2025-08-18T16:02:59.574195Z"}]}]},"status":"OPEN","goal_currency":"USD","start_date":"2025-08-15T11:28:27.794949Z","end_date":"2025-09-14T11:28:27.794822Z"}},"grant":"$undefined","doi":"10.55277/researchhub.lo09s2v9","topics":[{"id":7,"name":"Bioinformatics","slug":"$undefined","imageUrl":"$undefined","description":"$undefined","namespace":"$undefined","category":"$undefined","discussionCount":"$undefined","paperCount":"$undefined","subscriberCount":"$undefined","isLocked":"$undefined","isRemoved":"$undefined","isUsedForRep":"$undefined","raw":{"id":7,"name":"Bioinformatics"}},{"id":455,"name":"Protein Design","slug":"$undefined","imageUrl":"$undefined","description":"$undefined","namespace":"$undefined","category":"$undefined","discussionCount":"$undefined","paperCount":"$undefined","subscriberCount":"$undefined","isLocked":"$undefined","isRemoved":"$undefined","isUsedForRep":"$undefined","raw":{"id":455,"name":"Protein Design"}},{"id":61768,"name":"Cancer","slug":"$undefined","imageUrl":"$undefined","description":"$undefined","namespace":"$undefined","category":"$undefined","discussionCount":"$undefined","paperCount":"$undefined","subscriberCount":"$undefined","isLocked":"$undefined","isRemoved":"$undefined","isUsedForRep":"$undefined","raw":{"id":61768,"name":"Cancer"}}],"authors":[{"authorId":7944791,"userId":180556,"name":"Khondamir Rustamov","raw":{"id":7944791,"first_name":"Khondamir","last_name":"Rustamov","user":180556}}],"contacts":"$undefined","image":"https://storage.prod.researchhub.com/uploads/posts/users/180556/17935f37-2940-4dcf-9d43-3e1a933b5085/fig1-1.png","raw":{"id":4324,"status":"APPROVED","authors":["$21:props:children:props:topBanner:props:work:note:post:authors:0:raw"],"hubs":["$21:props:children:props:topBanner:props:work:note:post:topics:0:raw","$21:props:children:props:topBanner:props:work:note:post:topics:1:raw","$21:props:children:props:topBanner:props:work:note:post:topics:2:raw"],"unified_document":{"fundraise":"$21:props:children:props:topBanner:props:work:note:post:fundraise:raw","grant":null},"image_url":"https://storage.prod.researchhub.com/uploads/posts/users/180556/17935f37-2940-4dcf-9d43-3e1a933b5085/fig1-1.png","document_type":"PREREGISTRATION","slug":"foldcraft-an-open-framework-for-de-novo-design-of-egfr-targeting-nanobodies","doi":"10.55277/researchhub.lo09s2v9"}},"documentType":null,"proposalId":null,"registeredReportPrefill":null,"image":null,"previewImage":null,"topics":"$undefined","authors":"$undefined","raw":{"id":18252,"access":"WORKSPACE","latest_version":{"id":299113,"src":null,"created_date":"2025-08-15T11:28:10.542039Z","plain_text":"$57","json":"$58","note":18252},"organization":{"id":176241,"member_count":1,"user_permission":null,"created_date":"2025-08-04T08:33:59.915641Z","updated_date":"2025-08-04T08:33:59.915648Z","cover_image":null,"description":"","name":"Khondamir Rustamov's Org","note_created":false,"org_type":"PERSONAL","slug":"khondamir-rustamovs-org","user":180556},"post":"$21:props:children:props:topBanner:props:work:note:post:raw","registered_report_prefill":null,"unified_document":{"id":7555960,"is_removed":false},"created_date":"2025-08-15T11:19:07.561135Z","updated_date":"2025-08-15T11:24:22.303822Z","document_type":null,"title":"FoldCraft: An Open Framework for De Novo Design of EGFR-Targeting Nanobodies","created_by":180556},"content":null,"versionId":299113,"versionDate":"2025-08-15T11:28:10.542039Z","plainText":"$59","contentJson":"$5a"},"previewContent":"$5b","contentUrl":"https://storage.prod.researchhub.com/uploads/post_discussion/2025/08/15/RH-POST-PREREGISTRATION-USER-180556.txt","tips":[],"peerReviews":[{"id":10963,"createdBy":{"id":180918,"authorProfile":{"id":7964236,"fullName":"jaleel maulik","profileImage":"https://lh3.googleusercontent.com/a/ACg8ocJ3KcHBbLEDwmtmQpMm8dPBQCuiqEG-2b4fjo7azWN26IOByA=s96-c","isVerified":false}},"score":5,"createdDate":"2025-08-29T07:18:53.078926Z","contentPreview":"$undefined","isAssessed":false}],"image":"https://storage.prod.researchhub.com/uploads/posts/users/180556/17935f37-2940-4dcf-9d43-3e1a933b5085/fig1-1.png","enrichments":[],"moderationStatus":"APPROVED","raw":{"score":6,"user_flag":null,"authors":["$21:props:children:props:topBanner:props:work:authors:0:authorProfile:raw"],"id":4324,"created_by":{"id":180556,"author_profile":{"id":7944791,"is_public":true,"is_removed":false,"is_removed_date":null,"user":180556,"first_name":"Khondamir","last_name":"Rustamov","created_date":"2025-08-04T08:33:59.841980Z","updated_date":"2026-04-27T05:56:26.357541Z","description":"","h_index":3,"i10_index":0,"profile_image":"https://lh3.googleusercontent.com/a/ACg8ocI9wKqhqtic0gXOB7ULMtEvdp4_maQzczSa2lLNtFiFc-KH_4TRVA=s96-c","author_score":0,"university":null,"orcid_id":"https://orcid.org/0000-0001-7661-869X","openalex_ids":[],"education":[],"headline":"","facebook":null,"twitter":"","linkedin":"https://www.linkedin.com/in/khondamir-rustamov-840118180/","google_scholar":"https://scholar.google.com/citations?user=GofC7xEAAAAJ&hl=en","claimed":true,"country_code":null,"created_source":"RESEARCHHUB","last_full_fetch_from_openalex":null,"two_year_mean_citedness":2.4,"merged_with_author":null,"added_as_editor_date":null,"is_hub_editor_of":[],"is_hub_editor":false,"num_posts":2,"is_orcid_connected":true,"reputation":100,"reputation_v2":null,"reputation_list":[],"suspended_status":{"probable_spammer":false,"is_suspended":false},"total_score":null,"wallet":null,"is_verified":true},"balance":null,"balances":null,"created_date":"2025-08-04T08:33:59.820325Z","has_seen_first_coin_modal":false,"has_seen_orcid_connect_modal":false,"is_funder":false,"is_staking_opted_in":true,"is_suspended":false,"probable_spammer":false,"is_verified":true,"moderator":false,"reputation":100,"updated_date":"2025-08-05T07:39:14.830386Z","upload_tutorial_complete":false,"hub_rep":null,"time_rep":null},"created_date":"2025-08-15T11:28:27.721283Z","discussion_count":1,"document_type":"PREREGISTRATION","doi":"10.55277/researchhub.lo09s2v9","editor_type":"CK_EDITOR","full_markdown":"$5c","grants":[{"id":5,"short_title":"Translational Bioinformatics","status":"OPEN","organization":"ResearchHub Foundation","amount":"5000.00","currency":"USD","post_id":4233,"image_url":"https://storage.prod.researchhub.com/uploads/posts/users/39602/087ce0dd-8592-4569-b022-c38d1dea8964/chatgpt-image-jul-21-2025-11_43_16-am.png","title":"Request for Proposals - Bioinformatics & Genomics Research","applicant_count":28,"application_visibility":"OPTIONAL","proposal":{"unified_document_id":7555962,"ai_peer_review":{"id":80,"unified_document_id":7555962,"grant_id":5,"created_by_id":39602,"status":"completed","overall_rating":"good","overall_rationale":"The proposal presents a coherent, timely follow-up to validated computational work with clear deliverables (50 screened designs, open-source data release) and strong fit to the translational bioinformatics opportunity. The experimental plan is straightforward and outsourced to a credible partner (Adaptyv Bio), reducing execution risk. However, the proposal falls short of 'excellent' due to: (1) critical gaps in statistical analysis planning (no power analysis, no multiple-comparison strategy despite testing 50 designs); (2) limited evidence of the team's wet-lab capacity and institutional infrastructure for the proposed cancer cell-line validation; (3) modest publication track record (h-index 3-4, 7 works, 24 citations) that does not yet establish independent research leadership; and (4) vague institutional affiliation. The work is scientifically sound and addresses a real translational need, but the execution plan lacks sufficient methodological rigor and team-environment clarity to warrant a higher rating.","overall_score_numeric":2,"result_data":{"categories":{"methods_rigor":{"items":{"methods_detail":{"decision":"Partial","justification":"The in silico pipeline is described at a reasonable level (generate 500+ scaffolds, filter by AlphaFold3 metrics, select top 50 for screening). However, the functional validation methods are vague: 'standard cell-viability assays' in 'EGFR-positive cancer cell lines' lacks specificity on cell lines, controls, replicates, and statistical tests. The selection criteria for the top 50 are also underspecified."},"controls_defined":{"decision":"Partial","justification":"The proposal mentions that 'authenticated lines and mycoplasma-free status will be documented' for cell-line work, suggesting awareness of quality control. However, no explicit positive or negative controls for the binding assay or cell-viability assay are described. Adaptyv Bio presumably includes controls in their screening platform, but the proposal does not specify what they are."},"parameters_specified":{"decision":"Partial","justification":"Key parameters are specified for in silico filtering (pLDDT > 70, ipAE < 10, ipTM > 0.5) and for the initial design pool (500+ nanobodies, VHH fold). However, parameters for the functional validation (cell line passage number, viability assay type, positive/negative controls, replicates, statistical threshold) are not provided."},"model_choice_justified":{"decision":"Partial","justification":"The proposal justifies the use of AlphaFold2-Multimer for generative design (fold-conditioned hallucination) and AlphaFold3 for ranking (confidence metrics), but does not explain why AlphaFold3 metrics are expected to predict experimental binding better than AlphaFold2 metrics or other heuristics. The choice of SolMPNN for interface-preserving optimization is mentioned but not justified relative to alternatives."},"outcomes_linked_to_aims":{"decision":"Yes","justification":"The three hypotheses (H1: measurable binding; H2: AlphaFold3 metrics correlate with binding; H3: interface-preserving optimization improves binding) are directly linked to the stated aims. The experimental outcomes (binding affinity data, correlation analysis, functional validation) will directly test these hypotheses."}},"score":"Medium","rationale":"The in silico design pipeline is well-described (fold-conditioned hallucination, SolMPNN optimization, AlphaFold3 filtering) and builds on published methods. However, the proposal lacks sufficient detail on several critical aspects: (1) the selection criteria for the top 50 designs are vague ('AlphaFold3 metrics and diversity'); (2) controls for the experimental screening are not specified; (3) the functional validation in cancer cell lines is poorly detailed; and (4) the model choice (why AlphaFold2-Multimer for design, why AlphaFold3 for ranking) is not fully justified. The outcomes are linked to aims, but the methods for achieving them are incompletely specified."},"scientific_impact":{"items":{"generalizability":{"decision":"Partial","justification":"The project focuses on a single target (EGFR) and a single scaffold (VHH nanobodies), which limits generalizability. However, EGFR is a well-characterized, clinically relevant target, and the findings will likely inform design strategies for other targets. The open-source release of sequences and data will enable others to test generalizability."},"opens_new_directions":{"decision":"Yes","justification":"The project validates fold-conditioned hallucination as a design strategy and establishes a benchmark for AI-driven binder design. If successful, it will encourage adoption of AlphaFold-based design in the field and may inspire follow-up studies on other targets, scaffolds, or design strategies. The open-source data release will enable community-driven research."},"advances_understanding":{"decision":"Yes","justification":"The project directly tests whether FoldCraft's in silico success translates to experimental binding, addressing a critical gap in the field. It also tests whether AlphaFold3 confidence metrics (ipTM, pLDDT, iPAE) predict binding affinity, which is a novel empirical question with implications for structure-prediction-based design."}},"score":"High","rationale":"The project addresses a significant scientific challenge: whether AI-designed de novo nanobodies can translate from in silico predictions to experimental binding. This is a high-impact question in computational protein design and has direct implications for the field's ability to engineer therapeutic binders. The work advances understanding by testing whether AlphaFold3 confidence metrics predict real-world binding (a novel empirical question) and by releasing a large, open-source dataset of de novo nanobody sequences and affinities (rarely done in the field). The project opens new directions by validating fold-conditioned hallucination as a design strategy and by establishing a benchmark for future AI-driven binder-design tools."},"funding_opportunity_fit":{"items":{"fit_aims":{"decision":"Yes","justification":"The aims (validate FoldCraft designs experimentally, test whether AlphaFold3 metrics predict binding, release open-source data) are innovative and directly address a gap in de novo binder design. They align with the opportunity's focus on junior researchers pursuing innovative bioinformatics projects."},"fit_scope":{"decision":"Yes","justification":"The scope is well-defined and appropriately sized for a junior researcher with a $5,000 budget and 6-month timeline. The project builds on prior computational work (FoldCraft preprint, 2nd place in Adaptyv competition) and focuses on a single, well-characterized target (EGFR), making it feasible and focused."},"fit_modality":{"decision":"Yes","justification":"The proposal is fundamentally a bioinformatics project (AI-driven protein design using AlphaFold2-Multimer and SolMPNN) with an experimental validation component. This matches the translational bioinformatics scope."},"fit_deliverables":{"decision":"Yes","justification":"Deliverables include 50 experimentally screened nanobody designs, binding affinity data, correlation analysis between in silico and experimental metrics, and open-source release of sequences and models. These are concrete, measurable, and appropriate for a translational bioinformatics project."}},"score":"High","rationale":"The proposal aligns well with the ResearchHub Foundation's translational bioinformatics opportunity. FoldCraft is an innovative bioinformatics tool for protein design; the project pursues a junior researcher (Khondamir Rustamov, h-index 3-4) in a novel computational-experimental direction; and the work directly translates AI-designed sequences into experimental validation, bridging bioinformatics and wet-lab biology. The open-science commitment and focus on EGFR (a clinically relevant target) strengthen translational relevance."},"feasibility_and_execution":{"items":{"team_environment":{"decision":"Partial","justification":"The author (Khondamir Rustamov) has demonstrated computational protein-design expertise (FoldCraft preprint, 2nd place in Adaptyv competition, h-index 3-4). However, the institutional affiliation is unclear: public records list 'Center for Advanced Technologies' in Tashkent, Uzbekistan, which is not a major wet-lab antibody-engineering center. No co-investigators, collaborators, or mentors are mentioned. The proposal does not establish access to wet-lab facilities, cell-culture expertise, or institutional support for the functional validation component."},"ethics_data_quality":{"decision":"Partial","justification":"The proposal commits to open-sourcing all data and code, which supports reproducibility and data quality. Cell-line work will follow institutional biosafety policies, and authenticated, mycoplasma-free lines will be used. However, no IRB approval or ethics review is mentioned (though it may not be required for cell-line work). Data management and quality-assurance procedures for the Adaptyv Bio screening are not detailed; presumably Adaptyv Bio has internal QA, but the proposal does not specify how raw data will be validated or archived."},"procedures_feasible":{"decision":"Partial","justification":"The in silico design and Adaptyv Bio screening are feasible and well-established. However, the functional validation in cancer cell lines is presented as feasible but lacks detail: the proposal states it will use 'existing institutional resources and/or in-kind support' but does not specify the institution, lab, or PI. Without clear access to authenticated EGFR-positive cell lines, biosafety facilities, and expertise, this component is uncertain."},"timeline_milestones":{"decision":"Partial","justification":"The proposal states a 6-month timeline but does not provide detailed milestones. Step 1 (in silico design) and Step 2 (selection) are likely to take 1-2 months; Step 3 (Adaptyv Bio screening) typically takes 4-8 weeks; Step 4 (analysis) should take 2-4 weeks. However, the functional validation timeline is not specified, and it is unclear whether it will be completed within 6 months or extended. Milestones for data release and preprint submission are also not specified."},"recruitment_feasible":{"decision":"Yes","justification":"No human recruitment is required. The project involves computational design and outsourced experimental screening, so recruitment is not applicable. The proposal does not require IRB approval or human subjects."}},"score":"Medium","rationale":"The experimental screening component is highly feasible: Adaptyv Bio is a credible, established platform; the $5,000 budget is appropriate for 50 screens; and the 6-month timeline is realistic. Computational design is feasible on existing GPU infrastructure (RTX 3090/4080/4090). However, feasibility is compromised by: (1) unclear institutional affiliation and wet-lab capacity for the proposed cancer cell-line validation; (2) limited evidence of the team's prior experience with functional assays or cell-line work; (3) vague timeline and milestones for the functional validation component; and (4) modest track record (h-index 3-4) that does not establish independent research leadership."},"statistical_analysis_plan":{"items":{"power_analysis":{"decision":"No","justification":"No power analysis is provided. The proposal does not specify how many binders are expected to bind, what effect size is anticipated for the correlation between AlphaFold3 metrics and binding, or whether 50 designs are sufficient to detect meaningful correlations. This is a critical omission for a study with a fixed sample size."},"metrics_defined":{"decision":"Partial","justification":"Primary metrics are defined: AlphaFold3 confidence scores (ipTM, pLDDT, iPAE), binding affinity from Adaptyv Bio, and correlation coefficients. However, secondary metrics for functional validation (cell viability, IC50, antiproliferative effect) are not specified. The proposal does not define what constitutes a 'successful' binder or a 'strong' correlation."},"analysis_present":{"decision":"Partial","justification":"A basic analysis plan is outlined: compute correlations between AlphaFold3 metrics and experimental binding scores, visualize structures and rankings. However, the plan is superficial and does not address multiple-comparison correction, effect-size estimation, or confidence intervals."},"multiple_comparisons":{"decision":"No","justification":"No multiple-comparison correction strategy is mentioned. The proposal will test 50 designs and compute correlations between multiple AlphaFold3 metrics (ipTM, pLDDT, iPAE) and binding affinity, creating multiple statistical tests. Without correction (e.g., Bonferroni, FDR), false-positive rates will be inflated."},"analysis_matches_design":{"decision":"Partial","justification":"The proposed analysis (correlation between in silico metrics and experimental binding) matches the study design (rank 50 designs by AlphaFold3 metrics, screen experimentally, compare). However, the analysis plan does not account for the hierarchical structure of the data (multiple metrics per design, multiple designs per batch) or potential confounders (sequence similarity, epitope overlap)."}},"score":"Low","rationale":"The proposal lacks a rigorous statistical analysis plan. No power analysis is provided; no multiple-comparison correction strategy is specified despite testing 50 designs and computing correlations; no effect sizes or confidence intervals are mentioned; and the analysis section is limited to 'compute correlations between AlphaFold3 predictions and experimental scores' and 'visualize structures.' For a study testing 50 designs and computing multiple correlations, this is insufficient. The proposal does not address how many binders are expected to bind, what constitutes a 'successful' binder, or how false-discovery rate will be controlled."},"societal_and_broader_impact":{"items":{"societal_challenge":{"decision":"Yes","justification":"The project addresses cancer treatment, a major societal challenge. EGFR is implicated in multiple tumor types, and therapeutic binders could improve patient outcomes. The project also addresses a broader challenge in bioengineering: accelerating protein design through AI."},"commercial_potential":{"decision":"Partial","justification":"The project has potential commercial applications: validated de novo nanobodies against EGFR could be developed into therapeutics or research tools. However, the proposal does not discuss commercialization strategy, IP protection, or partnerships with industry. The open-source release may limit commercial potential but aligns with the proposal's commitment to open science."},"public_communication":{"decision":"Yes","justification":"The proposal commits to releasing all data, code, and sequences on GitHub and Zenodo with open CC-BY licensing, enabling public access and reuse. A preprint or publication will communicate findings to the scientific community. The open-source commitment supports transparency and reproducibility."}},"score":"High","rationale":"The project has strong societal and broader impact through its commitment to open science and data sharing. All sequences, models, and binding data will be released on GitHub and Zenodo under CC-BY licensing, creating a reusable resource for the scientific community. This democratizes access to AI-designed binders and enables independent validation and follow-up research. The project also addresses a societal challenge (cancer treatment) by advancing therapeutic binder design. The open-source commitment and focus on reproducibility align with broader trends in open science and responsible AI."},"clinical_or_translational_impact":{"items":{"unmet_need":{"decision":"Yes","justification":"The proposal addresses a clear unmet need: de novo design of therapeutic binders is slow, expensive, and dependent on animal immunization or brute-force screening. AI-driven design could accelerate development and reduce costs. EGFR is a well-characterized cancer target with clinical relevance."},"clinical_pathway":{"decision":"Partial","justification":"The proposal mentions functional validation in EGFR-positive cancer cell lines and antiproliferative effects, suggesting a path toward therapeutic development. However, the clinical pathway is not explicitly defined. The proposal does not specify whether the goal is to develop a lead candidate, establish a design platform, or generate a research tool. No discussion of lead optimization, toxicology, or regulatory pathways is provided."},"milestones_defined":{"decision":"Partial","justification":"The proposal defines milestones for the experimental screening (submit 50 designs, receive binding data) and data release (GitHub, Zenodo, preprint). However, milestones for translational development (lead optimization, functional validation, toxicology) are not specified. The functional validation in cancer cell lines is mentioned but not detailed with specific milestones or success criteria."}},"score":"Medium","rationale":"The project has clear translational potential: EGFR is a well-characterized cancer target; nanobodies are therapeutically relevant scaffolds; and the proposal includes functional validation in cancer cell lines. However, the clinical pathway is not well-defined. The proposal does not specify whether the goal is to develop a therapeutic candidate, establish a design platform for future therapeutics, or generate a research tool. The functional validation is presented as secondary and is poorly detailed. Milestones for translational development (e.g., lead optimization, toxicology, IND-enabling studies) are not defined. The project is better characterized as a proof-of-concept or platform validation than as a direct path to a clinical candidate."}},"fatal_flaws":[],"overall_rating":"good","major_strengths":["Strong computational validation and competitive track record: FoldCraft achieved 19.5% success rate in silico, outperformed state-of-the-art models (RFdiffusion, RFAntibody), and placed 2nd in Adaptyv Bio's EGFR binder competition with a validated binder (KD 4.77e-6 M), demonstrating genuine technical capability and external validation.","Clear translational relevance and unmet need: De novo binder design is a grand challenge in bioengineering; EGFR is a well-characterized cancer target; the proposal directly addresses whether AI-designed nanobodies can translate from in silico to experimental binding, filling a gap in the field.","Excellent open-science commitment and reproducibility: All sequences, AlphaFold3 models, binding data, analysis scripts, and the top-ranked design will be released on GitHub and Zenodo under CC-BY licensing, creating a reusable resource for the community and enabling independent validation.","Realistic scope and low-risk execution plan: Outsourcing screening to Adaptyv Bio (a credible, established platform) reduces technical risk; the $5,000 budget is modest and well-justified; 6-month timeline is achievable; no new infrastructure or ethics approvals required.","Well-motivated hypotheses and clear scientific questions: H1-H3 are testable and directly address whether in silico metrics predict real-world binding, whether AlphaFold3 confidence scores correlate with affinity, and whether interface-preserving optimization improves binding."],"overall_summary":"$5d","major_weaknesses":["Critical gap in statistical analysis planning: No power analysis, no multiple-comparison correction strategy despite testing 50 designs and computing correlations between AlphaFold3 metrics and binding scores. The proposal does not specify how many binders are expected to bind, what effect sizes are anticipated, or how false-discovery rate will be controlled.","Unclear institutional affiliation and wet-lab capacity: The proposal mentions 'existing institutional resources and/or in-kind support' for cancer cell-line validation but does not specify the institution, lab, or PI. The author's public affiliation (Center for Advanced Technologies, Tashkent, Uzbekistan) is not a major wet-lab antibody-engineering center, raising questions about access to authenticated cell lines, biosafety facilities, and expertise.","Limited independent research track record: Author has h-index 3-4, 7 total works, and 24 citations; most recent work is in 2024-2025 and appears to be preprints or early-stage publications. No evidence of prior independent funding, mentorship, or leadership of a research team, which is concerning for a junior researcher proposal.","Insufficient detail on functional validation methods: The proposal states that the top-ranked design will be tested in 'EGFR-positive cancer cell lines using standard cell-viability assays' but provides no detail on cell lines, controls, replicates, statistical tests, or expected effect sizes. This is presented as secondary but is critical for translational impact claims.","Vague selection criteria and potential bias: The proposal selects 50 designs based on 'AlphaFold3 confidence metrics and sequence/epitope diversity' but does not specify the weighting, clustering algorithm, or diversity metric. This could introduce selection bias and make results difficult to reproduce."],"overall_rationale":"The proposal presents a coherent, timely follow-up to validated computational work with clear deliverables (50 screened designs, open-source data release) and strong fit to the translational bioinformatics opportunity. The experimental plan is straightforward and outsourced to a credible partner (Adaptyv Bio), reducing execution risk. However, the proposal falls short of 'excellent' due to: (1) critical gaps in statistical analysis planning (no power analysis, no multiple-comparison strategy despite testing 50 designs); (2) limited evidence of the team's wet-lab capacity and institutional infrastructure for the proposed cancer cell-line validation; (3) modest publication track record (h-index 3-4, 7 works, 24 citations) that does not yet establish independent research leadership; and (4) vague institutional affiliation. The work is scientifically sound and addresses a real translational need, but the execution plan lacks sufficient methodological rigor and team-environment clarity to warrant a higher rating.","overall_confidence":"Medium","overall_score_numeric":2},"error_message":"","progress":100,"current_step":"Complete","llm_model":"us.anthropic.claude-haiku-4-5-20251001-v1:0","processing_time":56.48618658399937,"created_date":"2026-04-22T15:56:39.592181Z","updated_date":"2026-04-22T15:57:36.086434Z","editorial_feedback":null,"key_insight":null}}}],"hubs":[{"editor_permission_groups":[],"hub_image":"https://storage.prod.researchhub.com/uploads/hub_images/2020/10/07/bioinformatics.jpg","id":7,"is_locked":false,"is_removed":false,"name":"Bioinformatics","slug":"bioinformatics","namespace":"subcategory","is_used_for_rep":false},{"editor_permission_groups":[],"hub_image":null,"id":455,"is_locked":false,"is_removed":false,"name":"Protein Design","slug":"protein-design","namespace":null,"is_used_for_rep":false},{"editor_permission_groups":[],"hub_image":null,"id":61768,"is_locked":false,"is_removed":false,"name":"Cancer","slug":"cancer-1","namespace":null,"is_used_for_rep":false}],"image_url":"https://storage.prod.researchhub.com/uploads/posts/users/180556/17935f37-2940-4dcf-9d43-3e1a933b5085/fig1-1.png","is_latest_version":null,"is_removed":false,"is_root_version":true,"note":"$21:props:children:props:topBanner:props:work:note:raw","peer_reviews":[{"id":10963,"created_by":{"id":180918,"author_profile":{"id":7964236,"first_name":"jaleel","last_name":"maulik","profile_image":"https://lh3.googleusercontent.com/a/ACg8ocJ3KcHBbLEDwmtmQpMm8dPBQCuiqEG-2b4fjo7azWN26IOByA=s96-c"},"first_name":"jaleel","last_name":"maulik"},"created_date":"2025-08-29T07:18:53.078926Z","updated_date":"2025-08-29T07:18:53.078936Z","score":5,"is_assessed":false}],"post_src":"https://storage.prod.researchhub.com/uploads/post_discussion/2025/08/15/RH-POST-PREREGISTRATION-USER-180556.txt","preview_img":null,"registered_report_id":null,"renderable_text":"$5e","reviewed_by":null,"reviewed_date":null,"slug":"foldcraft-an-open-framework-for-de-novo-design-of-egfr-targeting-nanobodies","status":"APPROVED","title":"FoldCraft: An Open Framework for De Novo Design of EGFR-Targeting Nanobodies","unified_document_id":7555962,"unified_document":{"id":7555962,"documents":[{"id":4324,"title":"FoldCraft: An Open Framework for De Novo Design of EGFR-Targeting Nanobodies","slug":"foldcraft-an-open-framework-for-de-novo-design-of-egfr-targeting-nanobodies"}],"created_by":{"id":180556,"author_profile":{"id":7944791,"first_name":"Khondamir","last_name":"Rustamov","profile_image":"https://lh3.googleusercontent.com/a/ACg8ocI9wKqhqtic0gXOB7ULMtEvdp4_maQzczSa2lLNtFiFc-KH_4TRVA=s96-c"}},"reviews":{"avg":0,"count":0},"is_public":true,"is_removed":false,"document_type":"PREREGISTRATION"},"version_number":1,"updated_date":"2025-08-19T13:01:55.299933Z","bounty_type":null},"aiPeerReview":{"status":"completed","overallRating":"good","keyInsight":null,"raw":"$21:props:children:props:topBanner:props:work:raw:grants:0:proposal:ai_peer_review"},"linkedGrant":{"id":5,"postId":4233,"title":"Request for Proposals - Bioinformatics & Genomics Research","shortTitle":"Translational Bioinformatics","organization":"ResearchHub Foundation","fundingAmount":5000,"currency":"USD","status":"OPEN","applicationVisibility":"OPTIONAL","imageUrl":"https://storage.prod.researchhub.com/uploads/posts/users/39602/087ce0dd-8592-4569-b022-c38d1dea8964/chatgpt-image-jul-21-2025-11_43_16-am.png","applicantCount":28},"isPublic":true},"metadata":{"id":7555962,"score":4,"topics":[{"id":7,"name":"Bioinformatics","slug":"bioinformatics","imageUrl":"$undefined","description":"$undefined","namespace":"subcategory","category":"$undefined","discussionCount":"$undefined","paperCount":"$undefined","subscriberCount":"$undefined","isLocked":"$undefined","isRemoved":"$undefined","isUsedForRep":false,"raw":{"id":7,"name":"Bioinformatics","namespace":"subcategory","slug":"bioinformatics","created_date":"2019-11-27T02:50:12.611135Z","is_used_for_rep":false}},{"id":455,"name":"Protein Design","slug":"protein-design","imageUrl":"$undefined","description":"$undefined","namespace":null,"category":"$undefined","discussionCount":"$undefined","paperCount":"$undefined","subscriberCount":"$undefined","isLocked":"$undefined","isRemoved":"$undefined","isUsedForRep":false,"raw":{"id":455,"name":"Protein Design","namespace":null,"slug":"protein-design","created_date":"2023-09-29T16:21:02.902643Z","is_used_for_rep":false}},{"id":61768,"name":"Cancer","slug":"cancer-1","imageUrl":"$undefined","description":"$undefined","namespace":null,"category":"$undefined","discussionCount":"$undefined","paperCount":"$undefined","subscriberCount":"$undefined","isLocked":"$undefined","isRemoved":"$undefined","isUsedForRep":false,"raw":{"id":61768,"name":"Cancer","namespace":null,"slug":"cancer-1","created_date":"2023-11-06T22:50:00.193229Z","is_used_for_rep":false}}],"metrics":{"votes":4,"comments":1,"saves":0,"reviewScore":0,"conversationComments":1,"reviewComments":1,"bountyComments":1},"fundraising":{"id":105,"amountRaised":{"usd":0.64648207,"rsc":10},"goalAmount":{"usd":5000,"rsc":77341.66548501492},"contributors":{"numContributors":1,"topContributors":[{"id":31620,"authorProfile":{"id":928488,"fullName":"Ioana Gheorghe","firstName":"Ioana","lastName":"Gheorghe","profileImage":"https://storage.prod.researchhub.com/uploads/author_profile_images/2022/02/02/blob_xuIQVGV","headline":"","profileUrl":"/author/928488","user":"$undefined","description":"$undefined","createdDate":"$undefined","education":"$undefined","twitter":"$undefined","facebook":"$undefined","linkedin":"$undefined","googleScholar":"$undefined","orcidId":"$undefined","isOrcidConnected":false,"isClaimed":false,"hIndex":"$undefined","i10Index":"$undefined","userId":"$undefined","isVerified":false,"isHubEditor":false,"editorOfHubs":[],"raw":{"id":928488,"first_name":"Ioana","last_name":"Gheorghe","profile_image":"https://storage.prod.researchhub.com/uploads/author_profile_images/2022/02/02/blob_xuIQVGV"}},"totalContribution":{"rsc":10,"usd":0},"contributions":[{"amount":10,"currency":"RSC","date":"2025-08-18T16:02:59.574195Z"}]}]},"createdBy":{"id":0,"email":"","firstName":"","lastName":"","fullName":"Unknown User","isVerified":false,"authorProfile":"$undefined","balance":0,"lockedBalance":0,"totalRsc":0,"hasCompletedOnboarding":false,"moderator":false,"raw":null,"isModerator":false,"isFunder":false,"authProvider":"$undefined","isStakingOptedIn":false,"balanceHistory":0},"status":"OPEN","goalCurrency":"USD","startDate":"2025-08-15T11:28:27.794949Z","endDate":"2025-09-14T11:28:27.794822Z","createdDate":"","updatedDate":"","postId":"$undefined","postTitle":"$undefined","postSlug":"$undefined","postImage":null,"reviewMetrics":"$undefined","nonprofit":"$undefined","raw":{"id":105,"amount_raised":{"usd":0.64648207,"rsc":10},"goal_amount":{"usd":5000,"rsc":77341.66548501492},"contributors":{"total":1,"top":[{"id":31620,"author_profile":"$21:props:children:props:topBanner:props:metadata:fundraising:contributors:topContributors:0:authorProfile:raw","first_name":"Ioana","last_name":"Gheorghe","total_contribution":{"rsc":10,"rsc_usd_snapshot":5.664491989,"usd":0},"contributions":[{"amount":10,"currency":"RSC","date":"2025-08-18T16:02:59.574195Z"}]}]},"status":"OPEN","goal_currency":"USD","start_date":"2025-08-15T11:28:27.794949Z","end_date":"2025-09-14T11:28:27.794822Z"}},"bounties":[{"id":5445,"amount":"258.0000000000","status":"EXPIRED","expirationDate":"2026-08-10T23:39:19.828Z","bountyType":"BOUNTY","createdBy":{"id":39602,"email":"","firstName":"","lastName":"","fullName":"Unknown User","isVerified":false,"authorProfile":{"id":973869,"fullName":"ResearchHub Foundation","firstName":"ResearchHub","lastName":"Foundation","profileImage":"https://storage.prod.researchhub.com/uploads/author_profile_images/2025/05/28/blob_aBYdPow","headline":"","profileUrl":"/author/973869","user":"$undefined","description":"$undefined","createdDate":"$undefined","education":"$undefined","twitter":"$undefined","facebook":"$undefined","linkedin":"$undefined","googleScholar":"$undefined","orcidId":"$undefined","isOrcidConnected":false,"isClaimed":false,"hIndex":"$undefined","i10Index":"$undefined","userId":"$undefined","isVerified":false,"isHubEditor":false,"editorOfHubs":[],"raw":{"id":973869,"first_name":"ResearchHub","last_name":"Foundation","profile_image":"https://storage.prod.researchhub.com/uploads/author_profile_images/2025/05/28/blob_aBYdPow"}},"balance":0,"lockedBalance":0,"promotionalBalance":0,"fundingCredits":0,"balances":"$undefined","totalRsc":0,"hasCompletedOnboarding":false,"createdDate":"$undefined","moderator":false,"editorOfHubs":"$undefined","isModerator":false,"isFunder":false,"referralCode":"$undefined","authProvider":"$undefined","isStakingOptedIn":false,"balanceHistory":0,"raw":{"id":39602,"author_profile":"$21:props:children:props:topBanner:props:metadata:bounties:0:createdBy:authorProfile:raw"}},"solutions":[],"contributions":[],"totalAmount":"258","comment":"$undefined","raw":{"id":5445,"created_by":"$21:props:children:props:topBanner:props:metadata:bounties:0:createdBy:raw","amount":"258.0000000000","status":"EXPIRED"}}],"activeBounties":0,"closedBounties":0},"contentType":"post"}],"rightSidebar":"$L5f","children":"$L60"}]}]

FoldCraft: An Open Framework for De Novo Design of EGFR-Targeting Nanobodies

FoldCraft: An Open Framework for De Novo Design of EGFR-Targeting Nanobodies

Topics

DOI