Co-optimization of therapeutic antibody affinity and specificity using machine learning models that generalize to novel mutational space

Makowski, Emily K.; Kinnunen, Patrick C.; Jie, Huang; Wu, Lina; Smith, Matthew D.; Wang, Tiexin; Desai, Alec A.; Streu, Craig; Zhang, Yulei; Zupancic, Jennifer M.; Schardt, John S.; Linderman, Jennifer J.; Tessier, Peter M.

doi:10.1038/s41467-022-31457-3

Cited by 64 publications

(66 citation statements)

References 71 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Remarkably, for this particular task, using all of the models we consider, UniRep, ESM-1b, AntiBertY and AbLang exhibit performance inferior to one-hot encoding. This is consistent with results reported by Makowski et al, who found that UniRep or physicochemical properties did not improve performance for antibody affinity prediction compared with simple one-hot encoding 58 .…”

Section: Discussionsupporting

confidence: 93%

The RESP AI model accelerates the identification of tight-binding antibodies

2023

View full text Add to dashboard Cite

High-affinity antibodies are often identified through directed evolution, which may require many iterations of mutagenesis and selection to find an optimal candidate. Deep learning techniques hold the potential to accelerate this process but the existing methods cannot provide the confidence interval or uncertainty needed to assess the reliability of the predictions. Here we present a pipeline called RESP for efficient identification of high affinity antibodies. We develop a learned representation trained on over 3 million human B-cell receptor sequences to encode antibody sequences. We then develop a variational Bayesian neural network to perform ordinal regression on a set of the directed evolution sequences binned by off-rate and quantify their likelihood to be tight binders against an antigen. Importantly, this model can assess sequences not present in the directed evolution library and thus greatly expand the search space to uncover the best sequences for experimental evaluation. We demonstrate the power of this pipeline by achieving a 17-fold improvement in the KD of the PD-L1 antibody Atezolizumab and this success illustrates the potential of RESP in facilitating general antibody development.

show abstract

Section: Discussionsupporting

confidence: 93%

The RESP AI model accelerates the identification of tight-binding antibodies

2023

View full text Add to dashboard Cite

show abstract

“…As recently reported 24 , 51 , 52 , similar approaches could be applied to fully characterize sequence features of polyreactive conventional antibody clones. These methods can be expanded by analyzing large antigen-naïve libraries and adding in the three light-chain CDRs and germline gene choice as additional factors for polyreactivity prediction and optimization.…”

Section: Discussionmentioning

confidence: 81%

An in silico method to assess antibody fragment polyreactivity

et al. 2022

View full text Add to dashboard Cite

Antibodies are essential biological research tools and important therapeutic agents, but some exhibit non-specific binding to off-target proteins and other biomolecules. Such polyreactive antibodies compromise screening pipelines, lead to incorrect and irreproducible experimental results, and are generally intractable for clinical development. Here, we design a set of experiments using a diverse naïve synthetic camelid antibody fragment (nanobody) library to enable machine learning models to accurately assess polyreactivity from protein sequence (AUC > 0.8). Moreover, our models provide quantitative scoring metrics that predict the effect of amino acid substitutions on polyreactivity. We experimentally test our models’ performance on three independent nanobody scaffolds, where over 90% of predicted substitutions successfully reduced polyreactivity. Importantly, the models allow us to diminish the polyreactivity of an angiotensin II type I receptor antagonist nanobody, without compromising its functional properties. We provide a companion web-server that offers a straightforward means of predicting polyreactivity and polyreactivity-reducing mutations for any given nanobody sequence.

show abstract

“…A number of recent studies are demonstrating the potential of this field, for example machine learning-guided structural mutagenesis was used to improve the enantioselectivity and thermostability of a hydrolase enzyme used for degrading plastic waste (Lu et al ., 2022). Machine learning has been used in antibody engineering as well, such as predicting therapeutic antibody specificity for developability assessment (Mason et al ., 2021), off-target or polyspecific binding (Makowski et al ., 2022; Saksena et al ., 2022), and predicting escape of viral variants (e.g., SARS-CoV-2) to antibody drug candidates (Taft et al ., 2022). Furthermore, the de novo design of synthetic proteins via generative modeling has leveraged data driven deep learning to outperform physically based design methods on a variety of tasks including improving protein expression, stability, and ligand binding (Dauparas et al ., 2022; Wicky et al ., 2022; Shin et al ., 2021).…”

Section: Introductionmentioning

confidence: 99%

“…This enables the use of supervised machine learning methods, which rely on labeled data for training models. For example, display libraries of antibody variants can be screened for binding and non-binding to target antigen by fluorescence activated cell sorting (FACS) followed by deep sequencing, resulting in labeled data sets for supervised machine learning (Makowski et al ., 2022; Bryant et al ., 2021; Mason et al ., 2021). However, collecting high-quality, large, and well-labeled data sets requires extensive experimental workflows and can represent a bottleneck.…”

Section: Introductionmentioning

confidence: 99%

Meta Learning Improves Robustness and Performance in Machine Learning-Guided Protein Engineering

Minot

Reddy

2023

Preprint

View full text Add to dashboard Cite

Machine learning-guided protein engineering continues to rapidly progress, however, collecting large, well-labeled data sets remains time and resource intensive. Directed evolution and protein engineering studies often require extensive experimental processes to eliminate noise and fully label high-throughput protein sequence-function data. Meta learning methods established in other fields (e.g. computer vision and natural language processing) have proven effective in learning from noisy data, given the availability of a small data set with trusted labels and thus could be applied for protein engineering. Here, we generate yeast display antibody mutagenesis libraries and screen them for target antigen binding followed by deep sequencing. Meta learning approaches are able to learn under high synthetic and experimental noise as well as in under labeled data settings, typically outperforming baselines significantly and often requiring a fraction of the training data. Thus, we demonstrate meta learning may expedite and improve machine learning-guided protein engineering. Availability and implementation: The code used in this study is publicly available at https://github.com/LSSI-ETH/meta-learning-for-protein-engineering.

show abstract

Co-optimization of therapeutic antibody affinity and specificity using machine learning models that generalize to novel mutational space

Cited by 64 publications

References 71 publications

The RESP AI model accelerates the identification of tight-binding antibodies

The RESP AI model accelerates the identification of tight-binding antibodies

An in silico method to assess antibody fragment polyreactivity

Meta Learning Improves Robustness and Performance in Machine Learning-Guided Protein Engineering

Contact Info

Product

Resources

About