Deciphering the language of antibodies using self-supervised learning

Leem, Jinwoo; Mitchell, Laura; Farmery, James H.R.; Barton, Justin; Galson, Jacob D.

doi:10.1016/j.patter.2022.100513

Cited by 80 publications

(94 citation statements)

References 55 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The wealth of immune repertoire data provided by sequencing experiments has enabled development of antibody-specific language models. Models trained for masked language modeling have been shown to learn meaningful representations of immune repertoire sequences (21, 25, 26), and even repurposed to humanize antibodies (27). Generative models trained on sequence infilling have been shown to generate high-quality antibody libraries (28, 29).…”

Section: Introductionmentioning

confidence: 99%

Fast, accurate antibody structure prediction from deep learning on massive set of natural antibodies

Ruffolo

Chu

Mahajan

et al. 2022

Preprint

View full text Add to dashboard Cite

Antibodies have the capacity to bind a diverse set of antigens, and they have become critical therapeutics and diagnostic molecules. The binding of antibodies is facilitated by a set of six hypervariable loops that are diversified through genetic recombination and mutation. Even with recent advances, accurate structural prediction of these loops remains a challenge. Here, we present IgFold, a fast deep learning method for antibody structure prediction. IgFold consists of a pre-trained language model trained on 558M natural antibody sequences followed by graph networks that directly predict backbone atom coordinates. IgFold predicts structures of similar or better quality than alternative methods (including AlphaFold) in significantly less time (under one minute). Accurate structure prediction on this timescale makes possible avenues of investigation that were previously infeasible. As a demonstration of IgFold’s capabilities, we predicted structures for 105K paired antibody sequences, expanding the observed antibody structural space by over 40 fold.

show abstract

Section: Introductionmentioning

confidence: 99%

Fast, accurate antibody structure prediction from deep learning on massive set of natural antibodies

Ruffolo

Chu

Mahajan

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Examples of categorical predictions include predicting the source of an antibody (e.g. murine or human) from its amino acid sequence [ 24 ] or the incidence of amino acids at specific positions in protein sequence [ 25 ]. On the other hand, continuous predictions aim to capture values such as aggregation propensity [ 26 ] or orientation angles of residues in a structure [ 27 ].…”

Section: Encoding Antibody Antigen Sequence and Structure For Machine...mentioning

confidence: 99%

“…In such encoder–decoder architecture ( Figure 2E ), the network attempts to encode the input in a lower number of dimensions (encoder) and then reconstruct the original input from it (decoder). Several network architectures attempt that, such as variational autoencoders (VAEs) [ 34 ], Generative Adversarial Networks [ 35 ] or Transformers [ 25 ]. Latent representations can be trained from voluminous unlabeled datasets (e.g.…”

Section: Common Network Architectures Employed For Therapeutic Antibo...mentioning

confidence: 99%

“…This mechanism, therefore, allows selecting input elements that contribute more to better predictions and boost performance, especially in sequence-to-sequence tasks. Attention plays a crucial role in the Transformer model [ 38 ] and other state-of-the-art networks for text and sequence processing with recent antibody applications [ 25 , 39 ].…”

Section: Common Network Architectures Employed For Therapeutic Antibo...mentioning

confidence: 99%

“…Here one obscures (masks) part of the text and attempts to recreate it based on the learned context. There are currently three methods that were proposed for this problem, AntiBERTa [ 25 ], AntiBERTy [ 39 ] and AbLang [ 52 ].…”

Section: Embedding the Antibody Sequence Space: Applications Of Natur...mentioning

confidence: 99%

See 2 more Smart Citations

Machine-designed biotherapeutics: opportunities, feasibility and advantages of deep learning in computational antibody discovery

Wilman¹,

Wróbel²,

Bielska³

et al. 2022

Briefings in Bioinformatics

View full text Add to dashboard Cite

Antibodies are versatile molecular binders with an established and growing role as therapeutics. Computational approaches to developing and designing these molecules are being increasingly used to complement traditional lab-based processes. Nowadays, in silico methods fill multiple elements of the discovery stage, such as characterizing antibody–antigen interactions and identifying developability liabilities. Recently, computational methods tackling such problems have begun to follow machine learning paradigms, in many cases deep learning specifically. This paradigm shift offers improvements in established areas such as structure or binding prediction and opens up new possibilities such as language-based modeling of antibody repertoires or machine-learning-based generation of novel sequences. In this review, we critically examine the recent developments in (deep) machine learning approaches to therapeutic antibody design with implications for fully computational antibody design.

show abstract

Philosophical Problems of Immunology

Swiatczak

2024

Handbook of the Philosophy of Medicine

View full text Add to dashboard Cite

Deciphering the language of antibodies using self-supervised learning

Cited by 80 publications

References 55 publications

Fast, accurate antibody structure prediction from deep learning on massive set of natural antibodies

Fast, accurate antibody structure prediction from deep learning on massive set of natural antibodies

Machine-designed biotherapeutics: opportunities, feasibility and advantages of deep learning in computational antibody discovery

Philosophical Problems of Immunology

Contact Info

Product

Resources

About