Awesome antibody WIP🚧

• A curated list of machine learning(ML) & artificial intelligence(AI) methods for antibody. Inspired by Machine-learning-for-proteins.
• Warning⚠️: this list only covers a little part of computational antibody literature and heavily depends on my personal reading. CONTRIBUTIONS WELCOMED!

Categories

• Data
• Antibody structure prediction
• Antibody optimization and design
• Pretrained language model
• Properties prediction

Data

• Absolut!(2023)
  • A work form Victor Greiff group. Abosolut! allows the user to generate custom datasets by discretizing new antigens or annotating custom CDRH3 sequences enabling the on-demand generation of large synthetic datasets for 3D-antibody-antigen binding generated under the same, deterministic settings, with accessible computing resources. The documentation and code is available at GitHub.
  • Robert, Philippe A., Rahmad Akbar, Robert Frank, Milena Pavlović, Michael Widrich, Igor Snapkov, Andrei Slabodkin, et al. 2022. “Unconstrained Generation of Synthetic Antibody–Antigen Structures to Guide Machine Learning Methodology for Antibody Specificity Prediction.” Nature Computational Science 2 (12): 845–65. https://doi.org/10.1038/s43588-022-00372-4.
• Observed Antibody Space(2022)
  • Olsen, Tobias H., Fergus Boyles, and Charlotte M. Deane. 2022. “Observed Antibody Space: A Diverse Database of Cleaned, Annotated, and Translated Unpaired and Paired Antibody Sequences.” Protein Science 31 (1): 141–46. https://doi.org/10.1002/pro.4205.
  • The Observed Antibody Space (OAS) database was created in 2018 to offer clean, annotated, and translated repertoire data. The database is accessible at http://opig.stats.ox.ac.uk/webapps/oas/, and all data are freely available for download.
• Biophysical properties of the clinical-stage antibody landscape(2017)
  • Jain, Tushar, Tingwan Sun, Stéphanie Durand, Amy Hall, Nga Rewa Houston, Juergen H. Nett, Beth Sharkey, et al. 2017. “Biophysical Properties of the Clinical-Stage Antibody Landscape.” Proceedings of the National Academy of Sciences 114 (5): 944–49. https://doi.org/10.1073/pnas.1616408114.
  • A comprehensive analysis of these properties for essentially the full set of antibody drugs that have been tested in phase-2 or -3 clinical trials, or are approved by the FDA (as of 2017).

Antibody structure prediction

• IgFold(2022)
  • Currently SOTA of antibody structure prediction from sequence only. This work is from Jeffrey J. Gray lab using AntiBERTy pretrained language model. The code and plug-and-use python package is available: https://github.com/Graylab/IgFold.
  • Ruffolo, Jeffrey A., Lee-Shin Chu, Sai Pooja Mahajan, and Jeffrey J. Gray. 2022. “Fast, Accurate Antibody Structure Prediction from Deep Learning on Massive Set of Natural Antibodies.” bioRxiv. https://doi.org/10.1101/2022.04.20.488972.

Antibody optimization and design

• Fragment-based antibody de novo design(2022)
  • Aguilar Rangel, Mauricio, Alice Bedwell, Elisa Costanzi, Ross J. Taylor, Rosaria Russo, Gonçalo J. L. Bernardes, Stefano Ricagno, Judith Frydman, Michele Vendruscolo, and Pietro Sormanni. 2022. “Fragment-Based Computational Design of Antibodies Targeting Structured Epitopes.” Science Advances 8 (45): eabp9540. https://doi.org/10.1126/sciadv.abp9540.
  • Work form Pietro Sormanni. They grafted (antibody loop-like) fragments from known antibody structures and combined together to created a novel antibody binding to specific epitope on antigen, using crystal/predicted antigen structure. The experiment results shows fair binding affinity (nanomolar-level).
• Absci's works on antibody optimization(2022&2023)
  • Shanehsazzadeh, Amir, Sharrol Bachas, George Kasun, John M. Sutton, Andrea K. Steiger, Richard Shuai, Christa Kohnert, et al. 2023. “Unlocking de Novo Antibody Design with Generative Artificial Intelligence.” bioRxiv. https://doi.org/10.1101/2023.01.08.523187.
  • This work (by AI+antibody biotech company Absci) demonstrates that the approach of combining high-throughput affinity measurements and generative AI can accelerate drug(antibody) discovery. The data was made open-sourced.
  • A previous work of Absci on the same topic: Bachas, Sharrol, Goran Rakocevic, David Spencer, Anand V. Sastry, Robel Haile, John M. Sutton, George Kasun, et al. 2022. “Antibody Optimization Enabled by Artificial Intelligence Predictions of Binding Affinity and Naturalness.” bioRxiv. https://doi.org/10.1101/2022.08.16.504181.
• DeepAAI predict neutralizability (2022)
  • DeepAAI learns latent representation from known pairs of antibody-antigen. The representation was used to predict antibody's binding neutralization landscape to a set of antigens. The
  • Zhang, Jie, Yishan Du, Pengfei Zhou, Jinru Ding, Shuai Xia, Qian Wang, Feiyang Chen, et al. 2022. “Predicting Unseen Antibodies’ Neutralizability via Adaptive Graph Neural Networks.” Nature Machine Intelligence 4 (11): 964–76. https://doi.org/10.1038/s42256-022-00553-w.

Pretrained language model

• AntiBERTy(2021)
  • Ruffolo, Jeffrey A., Jeffrey J. Gray, and Jeremias Sulam. 2021. “Deciphering Antibody Affinity Maturation with Language Models and Weakly Supervised Learning.” arXiv. https://doi.org/10.48550/arXiv.2112.07782.
  • This work introduces AntiBERTy, a language model trained on 558M natural antibody sequences that finds that within repertoires, the model clusters antibodies into trajectories resembling affinity maturation, and shows that models trained to predict highly redundant sequences under a multiple instance learning framework identify key binding residues in the process.
• Ablang(2022)
  • Olsen, Tobias H, Iain H Moal, and Charlotte M Deane. 2022. “AbLang: An Antibody Language Model for Completing Antibody Sequences.” Bioinformatics Advances 2 (1): vbac046. https://doi.org/10.1093/bioadv/vbac046.
  • AbLang is a language model trained on the antibody sequences in the OAS database. AbLang restores the missing residues of antibody sequences better than using IMGT germlines or the general protein language model ESM-1b. AbLang does not require knowledge of the germline of the antibody and is seven times faster than ESM-1b. Abland is available as a python package.
• Antibody Benchmark(2022)
  • Published: Wang, Danqing, Fei Ye, and Hao Zhou. 2023. “On Pre-Trained Language Models for Antibody.” arXiv. https://doi.org/10.48550/arXiv.2301.12112.
  • OpenReview: Anonymous. 2022. “On Pre-Training Language Model for Antibody.” In . https://openreview.net/forum?id=zaq4LV55xHl.
  • This work develops the first comprehensive antibody benchmark, including tasks with varying antibody specificity: antigen-binding discrimination, antibody paratope prediction, B cell maturation analysis and antibody discovery. The authors observed that including biological mechanism of antibody maturation in the training process improve the language model's performance on highly antibody-specific question.

Properties prediction

• solPredict by Eli Lilly(2022)
  • Feng, Jiangyan, Min Jiang, James Shih, and Qing Chai. 2022. “Antibody Apparent Solubility Prediction from Sequence by Transfer Learning.” IScience 25 (10): 105173. https://doi.org/10.1016/j.isci.2022.105173.
  • Eli Lilly publish a research paper on iScience about constructing model solPredict for antibody apparent solubility prediction using protein pretrained model combined with transfer learning using in-house antibody solubility dataset. The code is open without dataset.
• Mutational landscape of SARS-COV2 RBD on ACE2/antibody binding(2022)
  • Taft, Joseph M., Cédric R. Weber, Beichen Gao, Roy A. Ehling, Jiami Han, Lester Frei, Sean W. Metcalfe, et al. 2022. “Deep Mutational Learning Predicts ACE2 Binding and Antibody Escape to Combinatorial Mutations in the SARS-CoV-2 Receptor-Binding Domain.” Cell 185 (21): 4008-4022.e14. https://doi.org/10.1016/j.cell.2022.08.024.
  • Selection and emergence of SARS CoV-2 variants are driven in part by mutations within the viral spike protein and in particular the ACE2 receptor-binding domain (RBD), a primary target site for neutralizing antibodies. Here, the authors develop deep mutational learning (DML), a machine-learning-guided protein engineering technology, which is used to investigate a massive sequence space of combinatorial mutations, representing billions of RBD variants, by accurately predicting their impact on ACE2 binding and antibody escape.

• Online platform for antibody resarch

• BioPhi by Merck(2022)
  • Prihoda, David, Jad Maamary, Andrew Waight, Veronica Juan, Laurence Fayadat-Dilman, Daniel Svozil, and Danny A. Bitton. 2022. “BioPhi: A Platform for Antibody Design, Humanization, and Humanness Evaluation Based on Natural Antibody Repertoires and Deep Learning.” MAbs 14 (1): 2020203. https://doi.org/10.1080/19420862.2021.2020203.
  • An open-sourced platform named BioPhi by Merck and researchers from academia. BioPhi is consist of two models: Sapiens is a deep learning humanization method trained on the OAS using language modeling; OASis is a granular, interpretable and diverse humanness score based on 9-mer peptide search in the OAS. BioPhi thus offers an antibody design interface with automated methods that capture the richness of natural antibody repertoires to produce therapeutics with desired properties and accelerate antibody discovery campaigns. The BioPhi platform is accessible at https://biophi.dichlab.org and https://github.com/Merck/BioPhi.