Deep generative models of genetic variation capture the effects of mutations

ProteinGym: Large-Scale Benchmarks for Protein Design and Fitness Prediction

Predicting the effects of mutations in proteins is critical to many applications, from understanding genetic disease to designing novel proteins that can address our most pressing challenges in climate, agriculture and healthcare. Despite a surge in machine learning-based protein models to tackle these questions, an assessment of their respective benefits is challenging due to the use of distinct, often contrived, experimental datasets, and the variable performance of models across different protein families. Addressing these challenges requires scale. To that end we introduce ProteinGym, a large-scale and holistic set of benchmarks specifically designed for protein fitness prediction and design. It encompasses both a broad collection of over 250 standardized deep mutational scanning assays, spanning millions of mutated sequences, as well as curated clinical datasets providing high-quality expert annotations about mutation effects. We devise a robust evaluation framework that combines metrics for both fitness prediction and design, factors in known limitations of the underlying experimental methods, and covers both zero-shot and supervised settings. We report the performance of a diverse set of over 70 high-performing models from various subfields (eg., alignment-based, inverse folding) into a unified benchmark suite. We open source the corresponding codebase, datasets, MSAs, structures, model predictions and develop a user-friendly website that facilitates data access and analysis.

ProteinNPT: Improving Protein Property Prediction and Design with Non-Parametric Transformers

Protein design holds immense potential for optimizing naturally occurring proteins, with broad applications in drug discovery, material design, and sustainability. However, computational methods for protein engineering are confronted with significant challenges, such as an expansive design space, sparse functional regions, and a scarcity of available labels. These issues are further exacerbated in practice by the fact most real-life design scenarios necessitate the simultaneous optimization of multiple properties. In this work, we introduce ProteinNPT, a non-parametric transformer variant tailored to protein sequences and particularly suited to label-scarce and multi-task learning settings. We first focus on the supervised fitness prediction setting and develop several cross-validation schemes which support robust performance assessment. We subsequently reimplement prior top-performing baselines, introduce several extensions of these baselines by integrating diverse branches of the protein engineering literature, and demonstrate that ProteinNPT consistently outperforms all of them across a diverse set of protein property prediction tasks. Finally, we demonstrate the value of our approach for iterative protein design across extensive in silico Bayesian optimization and conditional sampling experiments.

Deep generative models of LDLR protein structure to predict variant pathogenicity

The complex structure and function of low density lipoprotein receptor (LDLR) makes classification of protein-coding missense variants challenging. Deep generative models, including Evolutionary model of Variant Effect (EVE), Evolutionary Scale Modeling (ESM), and AlphaFold 2 (AF2), have enabled significant progress in the prediction of protein structure and function. ESM and EVE directly estimate the likelihood of a variant sequence but are purely data-driven and challenging to interpret. AF2 predicts LDLR structures, but variant effects are explicitly modeled by estimating changes in stability. We tested the effectiveness of these models for predicting variant pathogenicity compared to established methods. AF2 produced two distinct conformations based on a novel hinge mechanism. Within ESM's hidden space, benign and pathogenic variants had different distributions. In EVE, these distributions were similar. EVE and ESM were comparable to Polyphen-2, SIFT, REVEL, and Primate AI for predicting binary classifications in ClinVar. However, they were more strongly correlated with experimental measures of LDL uptake. AF2 poorly performed in these tasks. Using the UK Biobank to compare association with clinical phenotypes, ESM and EVE were more strongly associated with serum LDL-C than Polyphen-2. ESM was able to identify variants with more extreme LDL-C levels than EVE and had a significantly stronger association with atherosclerotic cardiovascular disease. In conclusion, AF2 predicted LDLR structures do not accurately model variant pathogenicity. ESM and EVE are competitive with prior scoring methods for prediction based on binary classifications in ClinVar but are superior based on correlations with experimental assays and clinical phenotypes.

Harnessing generative AI to decode enzyme catalysis and evolution for enhanced engineering

Enzymes, as paramount protein catalysts, occupy a central role in fostering remarkable progress across numerous fields. However, the intricacy of sequence-function relationships continues to obscure our grasp of enzyme behaviors and curtails our capabilities in rational enzyme engineering. Generative artificial intelligence (AI), known for its proficiency in handling intricate data distributions, holds the potential to offer novel perspectives in enzyme research. Generative models could discern elusive patterns within the vast sequence space and uncover new functional enzyme sequences. This review highlights the recent advancements in employing generative AI for enzyme sequence analysis. We delve into the impact of generative AI in predicting mutation effects on enzyme fitness, catalytic activity and stability, rationalizing the laboratory evolution of de novo enzymes, and decoding protein sequence semantics and their application in enzyme engineering. Notably, the prediction of catalytic activity and stability of enzymes using natural protein sequences serves as a vital link, indicating how enzyme catalysis shapes enzyme evolution. Overall, we foresee that the integration of generative AI into enzyme studies will remarkably enhance our knowledge of enzymes and expedite the creation of superior biocatalysts.

Enhancing luciferase activity and stability through generative modeling of natural enzyme sequences

The availability of natural protein sequences synergized with generative AI provides new paradigms to engineer enzymes. Although active enzyme variants with numerous mutations have been designed using generative models, their performance often falls short of their wild type counterparts. Additionally, in practical applications, choosing fewer mutations that can rival the efficacy of extensive sequence alterations is usually more advantageous. Pinpointing beneficial single mutations continues to be a formidable task. In this study, using the generative maximum entropy model to analyze Renilla luciferase (RLuc) homologs, and in conjunction with biochemistry experiments, we demonstrated that natural evolutionary information could be used to predictively improve enzyme activity and stability by engineering the active center and protein scaffold, respectively. The success rate to improve either luciferase activity or stability of designed single mutants is ~50%. This finding highlights nature's ingenious approach to evolving proficient enzymes, wherein diverse evolutionary pressures are preferentially applied to distinct regions of the enzyme, ultimately culminating in an overall high performance. We also reveal an evolutionary preference in RLuc toward emitting blue light that holds advantages in terms of water penetration compared to other light spectra. Taken together, our approach facilitates navigation through enzyme sequence space and offers effective strategies for computer-aided rational enzyme engineering.

Unsupervisedly Prompting AlphaFold2 for Accurate Few-Shot Protein Structure Prediction

Data-driven predictive methods that can efficiently and accurately transform protein sequences into biologically active structures are highly valuable for scientific research and medical development. Determining an accurate folding landscape using coevolutionary information is fundamental to the success of modern protein structure prediction methods. As the state of the art, AlphaFold2 has dramatically raised the accuracy without performing explicit coevolutionary analysis. Nevertheless, its performance still shows strong dependence on available sequence homologues. Based on the interrogation on the cause of such dependence, we presented EvoGen, a meta generative model, to remedy the underperformance of AlphaFold2 for poor MSA targets. By prompting the model with calibrated or virtually generated homologue sequences, EvoGen helps AlphaFold2 fold accurately in the low-data regime and even achieve encouraging performance with single-sequence predictions. Being able to make accurate predictions with few-shot MSA not only generalizes AlphaFold2 better for orphan sequences but also democratizes its use for high-throughput applications. Besides, EvoGen combined with AlphaFold2 yields a probabilistic structure generation method that could explore alternative conformations of protein sequences, and the task-aware differentiable algorithm for sequence generation will benefit other related tasks including protein design.

Coherent Blending of Biophysics-Based Knowledge with Bayesian Neural Networks for Robust Protein Property Prediction

Predicting properties of proteins is of interest for basic biological understanding and protein engineering alike. Increasingly, machine learning (ML) approaches are being used for this task. However, the accuracy of such ML models typically degrades as test proteins stray further from the training data distribution. On the other hand, models that are more data-free, such as biophysics-based models, are typically uniformly accurate over all of the protein space, even if inferior for test points close to the training distribution. Consequently, being able to cohesively blend these two types of information within one model, as appropriate in different parts of the protein space, will improve overall importance. Herein, we tackle just this problem to yield a simple, practical, and scalable approach that can be easily implemented. In particular, we use a Bayesian formulation to integrate biophysical knowledge into neural networks. However, in doing so, a technical challenge arises: Bayesian neural networks (BNNs) enable the user to specify prior information only on the neural network weight parameters, rather than on the function values given to us from a typical biophysics-based model. Consequently, we devise a principled probabilistic method to overcome this challenge. Our approach yields intuitively pleasing results: predictions rely more heavily on the biophysical prior information when the BNN epistemic uncertainty─uncertainty arising from a lack of training data rather than sensor noise─is large and more heavily on the neural network when the epistemic uncertainty is small. We demonstrate this approach on an illustrative synthetic example, on two examples of protein property prediction (fluorescence and binding), and for generality on one small molecule property prediction problem.

ProGen2: Exploring the boundaries of protein language models

Attention-based models trained on protein sequences have demonstrated incredible success at classification and generation tasks relevant for artificial-intelligence-driven protein design. However, we lack a sufficient understanding of how very large-scale models and data play a role in effective protein model development. We introduce a suite of protein language models, named ProGen2, that are scaled up to 6.4B parameters and trained on different sequence datasets drawn from over a billion proteins from genomic, metagenomic, and immune repertoire databases. ProGen2 models show state-of-the-art performance in capturing the distribution of observed evolutionary sequences, generating novel viable sequences, and predicting protein fitness without additional fine-tuning. As large model sizes and raw numbers of protein sequences continue to become more widely accessible, our results suggest that a growing emphasis needs to be placed on the data distribution provided to a protein sequence model. Our models and code are open sourced for widespread adoption in protein engineering. A record of this paper's Transparent Peer Review process is included in the supplemental information.

A new age in protein design empowered by deep learning

The rapid progress in the field of deep learning has had a significant impact on protein design. Deep learning methods have recently produced a breakthrough in protein structure prediction, leading to the availability of high-quality models for millions of proteins. Along with novel architectures for generative modeling and sequence analysis, they have revolutionized the protein design field in the past few years remarkably by improving the accuracy and ability to identify novel protein sequences and structures. Deep neural networks can now learn and extract the fundamental features of protein structures, predict how they interact with other biomolecules, and have the potential to create new effective drugs for treating disease. As their applicability in protein design is rapidly growing, we review the recent developments and technology in deep learning methods and provide examples of their performance to generate novel functional proteins.

Continuous evolution of user-defined genes at 1-million-times the genomic mutation rate

When nature maintains or evolves a gene's function over millions of years at scale, it produces a diversity of homologous sequences whose patterns of conservation and change contain rich structural, functional, and historical information about the gene. However, natural gene diversity likely excludes vast regions of functional sequence space and includes phylogenetic and evolutionary eccentricities, limiting what information we can extract. We introduce an accessible experimental approach for compressing long-term gene evolution to laboratory timescales, allowing for the direct observation of extensive adaptation and divergence followed by inference of structural, functional, and environmental constraints for any selectable gene. To enable this approach, we developed a new orthogonal DNA replication (OrthoRep) system that durably hypermutates chosen genes at a rate of >10 -4 substitutions per base in vivo . When OrthoRep was used to evolve a conditionally essential maladapted enzyme, we obtained thousands of unique multi-mutation sequences with many pairs >60 amino acids apart (>15% divergence), revealing known and new factors influencing enzyme adaptation. The fitness of evolved sequences was not predictable by advanced machine learning models trained on natural variation. We suggest that OrthoRep supports the prospective and systematic discovery of constraints shaping gene evolution, uncovering of new regions in fitness landscapes, and general applications in biomolecular engineering.

Accurate prediction of functional effect of single amino acid variants with deep learning

The assessment of functional effect of amino acid variants is a critical biological problem in proteomics for clinical medicine and protein engineering. Although natively occurring variants offer insights into deleterious variants, high-throughput deep mutational experiments enable comprehensive investigation of amino acid variants for a given protein. However, these mutational experiments are too expensive to dissect millions of variants on thousands of proteins. Thus, computational approaches have been proposed, but they heavily rely on hand-crafted evolutionary conservation, limiting their accuracy. Recent advancement in transformers provides a promising solution to precisely estimate the functional effects of protein variants on high-throughput experimental data. Here, we introduce a novel deep learning model, namely Rep2Mut-V2, which leverages learned representation from transformer models. Rep2Mut-V2 significantly enhances the prediction accuracy for 27 types of measurements of functional effects of protein variants. In the evaluation of 38 protein datasets with 118,933 single amino acid variants, Rep2Mut-V2 achieved an average Spearman's correlation coefficient of 0.7. This surpasses the performance of six state-of-the-art methods, including the recently released methods ESM, DeepSequence and EVE. Even with limited training data, Rep2Mut-V2 outperforms ESM and DeepSequence, showing its potential to extend high-throughput experimental analysis for more protein variants to reduce experimental cost. In conclusion, Rep2Mut-V2 provides accurate predictions of the functional effects of single amino acid variants of protein coding sequences. This tool can significantly aid in the interpretation of variants in human disease studies.

Alignment-based Protein Mutational Landscape Prediction: Doing More with Less

The wealth of genomic data has boosted the development of computational methods predicting the phenotypic outcomes of missense variants. The most accurate ones exploit multiple sequence alignments, which can be costly to generate. Recent efforts for democratizing protein structure prediction have overcome this bottleneck by leveraging the fast homology search of MMseqs2. Here, we show the usefulness of this strategy for mutational outcome prediction through a large-scale assessment of 1.5M missense variants across 72 protein families. Our study demonstrates the feasibility of producing alignment-based mutational landscape predictions that are both high-quality and compute-efficient for entire proteomes. We provide the community with the whole human proteome mutational landscape and simplified access to our predictive pipeline.

GENERALIST: A latent space based generative model for protein sequence families

Generative models of protein sequence families are an important tool in the repertoire of protein scientists and engineers alike. However, state-of-the-art generative approaches face inference, accuracy, and overfitting- related obstacles when modeling moderately sized to large proteins and/or protein families with low sequence coverage. Here, we present a simple to learn, tunable, and accurate generative model, GENERALIST: GENERAtive nonLInear tenSor-factorizaTion for protein sequences. GENERALIST accurately captures several high order summary statistics of amino acid covariation. GENERALIST also predicts conservative local optimal sequences which are likely to fold in stable 3D structure. Importantly, unlike current methods, the density of sequences in GENERALIST-modeled sequence ensembles closely resembles the corresponding natural ensembles. Finally, GENERALIST embeds protein sequences in an informative latent space. GENERALIST will be an important tool to study protein sequence variability.

Zero-shot mutation effect prediction on protein stability and function using RoseTTAFold

Predicting the effects of mutations on protein function and stability is an outstanding challenge. Here, we assess the performance of a variant of RoseTTAFold jointly trained for sequence and structure recovery, RFjoint , for mutation effect prediction. Without any further training, we achieve comparable accuracy in predicting mutation effects for a diverse set of protein families using RFjoint to both another zero-shot model (MSA Transformer) and a model that requires specific training on a particular protein family for mutation effect prediction (DeepSequence). Thus, although the architecture of RFjoint was developed to address the protein design problem of scaffolding functional motifs, RFjoint acquired an understanding of the mutational landscapes of proteins during model training that is equivalent to that of recently developed large protein language models. The ability to simultaneously reason over protein structure and sequence could enable even more precise mutation effect predictions following supervised training on the task. These results suggest that RFjoint has a quite broad understanding of protein sequence-structure landscapes, and can be viewed as a joint model for protein sequence and structure which could be broadly useful for protein modeling.

Illuminating protein space with a programmable generative model

Three billion years of evolution has produced a tremendous diversity of protein molecules1, but the full potential of proteins is likely to be much greater. Accessing this potential has been challenging for both computation and experiments because the space of possible protein molecules is much larger than the space of those likely to have functions. Here we introduce Chroma, a generative model for proteins and protein complexes that can directly sample novel protein structures and sequences, and that can be conditioned to steer the generative process towards desired properties and functions. To enable this, we introduce a diffusion process that respects the conformational statistics of polymer ensembles, an efficient neural architecture for molecular systems that enables long-range reasoning with sub-quadratic scaling, layers for efficiently synthesizing three-dimensional structures of proteins from predicted inter-residue geometries and a general low-temperature sampling algorithm for diffusion models. Chroma achieves protein design as Bayesian inference under external constraints, which can involve symmetries, substructure, shape, semantics and even natural-language prompts. The experimental characterization of 310 proteins shows that sampling from Chroma results in proteins that are highly expressed, fold and have favourable biophysical properties. The crystal structures of two designed proteins exhibit atomistic agreement with Chroma samples (a backbone root-mean-square deviation of around 1.0 Å). With this unified approach to protein design, we hope to accelerate the programming of protein matter to benefit human health, materials science and synthetic biology.

Exploring the Protein Sequence Space with Global Generative Models

Recent advancements in specialized large-scale architectures for training images and language have profoundly impacted the field of computer vision and natural language processing (NLP). Language models, such as the recent ChatGPT and GPT-4, have demonstrated exceptional capabilities in processing, translating, and generating human language. These breakthroughs have also been reflected in protein research, leading to the rapid development of numerous new methods in a short time, with unprecedented performance. Several of these models have been developed with the goal of generating sequences in novel regions of the protein space. In this work, we provide an overview of the use of protein generative models, reviewing (1) language models for the design of novel artificial proteins, (2) works that use non-transformer architectures, and (3) applications in directed evolution approaches.

Comparative genomics hints at dispensability of multiple essential genes in two Escherichia coli L-form strains

Despite the critical role of bacterial cell walls in maintaining cell shapes, certain environmental stressors can induce the transition of many bacterial species into a wall-deficient state called L-form. Long-term induced Escherichia coli L-forms lose their rod shape and usually hold significant mutations that affect cell division and growth. Besides this, the genetic background of L-form bacteria is still poorly understood. In the present study, the genomes of two stable L-form strains of E. coli (NC-7 and LWF+) were sequenced and their gene mutation status was determined and compared with their parental strains. Comparative genomic analysis between two L-forms reveals both unique adaptions and common mutated genes, many of which belong to essential gene categories not involved in cell wall biosynthesis, indicating that L-form genetic adaptation impacts crucial metabolic pathways. Missense variants from L-forms and Lenski's long-term evolution experiment (LTEE) were analyzed in parallel using an optimized DeepSequence pipeline to investigate predicted mutation effects (α) on protein functions. We report that the two L-form strains analyzed display a frequency of 6-10% (0% for LTEE) in mutated essential genes where the missense variants have substantial impact on protein functions (α<0.5). This indicates the emergence of different survival strategies in L-forms through changes in essential genes during adaptions to cell wall deficiency. Collectively, our results shed light on the detailed genetic background of two E. coli L-forms and pave the way for further investigations of the gene functions in L-form bacterial models.

Harnessing Generative AI to Decode Enzyme Catalysis and Evolution for Enhanced Engineering

Enzymes, as paramount protein catalysts, occupy a central role in fostering remarkable progress across numerous fields. However, the intricacy of sequence-function relationships continues to obscure our grasp of enzyme behaviors and curtails our capabilities in rational enzyme engineering. Generative artificial intelligence (AI), known for its proficiency in handling intricate data distributions, holds the potential to offer novel perspectives in enzyme research. By applying generative models, we could discern elusive patterns within the vast sequence space and uncover new functional enzyme sequences. This review highlights the recent advancements in employing generative AI for enzyme sequence analysis. We delve into the impact of generative AI in predicting mutation effects on enzyme fitness, activity, and stability, rationalizing the laboratory evolution of de novo enzymes, decoding protein sequence semantics, and its applications in enzyme engineering. Notably, the prediction of enzyme activity and stability using natural enzyme sequences serves as a vital link, indicating how enzyme catalysis shapes enzyme evolution. Overall, we foresee that the integration of generative AI into enzyme studies will remarkably enhance our knowledge of enzymes and expedite the creation of superior biocatalysts.

Enhancing Luciferase Activity and Stability through Generative Modeling of Natural Enzyme Sequences

The availability of natural protein sequences synergized with generative artificial intelligence (AI) provides new paradigms to create enzymes. Although active enzyme variants with numerous mutations have been produced using generative models, their performance often falls short compared to their wild-type counterparts. Additionally, in practical applications, choosing fewer mutations that can rival the efficacy of extensive sequence alterations is usually more advantageous. Pinpointing beneficial single mutations continues to be a formidable task. In this study, using the generative maximum entropy model to analyze Renilla luciferase homologs, and in conjunction with biochemistry experiments, we demonstrated that natural evolutionary information could be used to predictively improve enzyme activity and stability by engineering the active center and protein scaffold, respectively. The success rate of designed single mutants is ~50% to improve either luciferase activity or stability. These finding highlights nature's ingenious approach to evolving proficient enzymes, wherein diverse evolutionary pressures are preferentially applied to distinct regions of the enzyme, ultimately culminating in an overall high performance. We also reveal an evolutionary preference in Renilla luciferase towards emitting blue light that holds advantages in terms of water penetration compared to other light spectrum. Taken together, our approach facilitates navigation through enzyme sequence space and offers effective strategies for computer-aided rational enzyme engineering.

Learning from prepandemic data to forecast viral escape

Effective pandemic preparedness relies on anticipating viral mutations that are able to evade host immune responses to facilitate vaccine and therapeutic design. However, current strategies for viral evolution prediction are not available early in a pandemic-experimental approaches require host polyclonal antibodies to test against1-16, and existing computational methods draw heavily from current strain prevalence to make reliable predictions of variants of concern17-19. To address this, we developed EVEscape, a generalizable modular framework that combines fitness predictions from a deep learning model of historical sequences with biophysical and structural information. EVEscape quantifies the viral escape potential of mutations at scale and has the advantage of being applicable before surveillance sequencing, experimental scans or three-dimensional structures of antibody complexes are available. We demonstrate that EVEscape, trained on sequences available before 2020, is as accurate as high-throughput experimental scans at anticipating pandemic variation for SARS-CoV-2 and is generalizable to other viruses including influenza, HIV and understudied viruses with pandemic potential such as Lassa and Nipah. We provide continually revised escape scores for all current strains of SARS-CoV-2 and predict probable further mutations to forecast emerging strains as a tool for continuing vaccine development ( evescape.org ).

Infer global, predict local: Quantity-relevance trade-off in protein fitness predictions from sequence data

Predicting the effects of mutations on protein function is an important issue in evolutionary biology and biomedical applications. Computational approaches, ranging from graphical models to deep-learning architectures, can capture the statistical properties of sequence data and predict the outcome of high-throughput mutagenesis experiments probing the fitness landscape around some wild-type protein. However, how the complexity of the models and the characteristics of the data combine to determine the predictive performance remains unclear. Here, based on a theoretical analysis of the prediction error, we propose descriptors of the sequence data, characterizing their quantity and relevance relative to the model. Our theoretical framework identifies a trade-off between these two quantities, and determines the optimal subset of data for the prediction task, showing that simple models can outperform complex ones when inferred from adequately-selected sequences. We also show how repeated subsampling of the sequence data is informative about how much epistasis in the fitness landscape is not captured by the computational model. Our approach is illustrated on several protein families, as well as on in silico solvable protein models.

Artificial intelligence-aided protein engineering: from topological data analysis to deep protein language models

Protein engineering is an emerging field in biotechnology that has the potential to revolutionize various areas, such as antibody design, drug discovery, food security, ecology, and more. However, the mutational space involved is too vast to be handled through experimental means alone. Leveraging accumulative protein databases, machine learning (ML) models, particularly those based on natural language processing (NLP), have considerably expedited protein engineering. Moreover, advances in topological data analysis (TDA) and artificial intelligence-based protein structure prediction, such as AlphaFold2, have made more powerful structure-based ML-assisted protein engineering strategies possible. This review aims to offer a comprehensive, systematic, and indispensable set of methodological components, including TDA and NLP, for protein engineering and to facilitate their future development.

ETLD: an encoder-transformation layer-decoder architecture for protein contact and mutation effects prediction

The latent features extracted from the multiple sequence alignments (MSAs) of homologous protein families are useful for identifying residue-residue contacts, predicting mutation effects, shaping protein evolution, etc. Over the past three decades, a growing body of supervised and unsupervised machine learning methods have been applied to this field, yielding fruitful results. Here, we propose a novel self-supervised model, called encoder-transformation layer-decoder (ETLD) architecture, capable of capturing protein sequence latent features directly from MSAs. Compared to the typical autoencoder model, ETLD introduces a transformation layer with the ability to learn inter-site couplings, which can be used to parse out the two-dimensional residue-residue contacts map after a simple mathematical derivation or an additional supervised neural network. ETLD retains the process of encoding and decoding sequences, and the predicted probabilities of amino acids at each site can be further used to construct the mutation landscapes for mutation effects prediction, outperforming advanced models such as GEMME, DeepSequence and EVmutation in general. Overall, ETLD is a highly interpretable unsupervised model with great potential for improvement and can be further combined with supervised methods for more extensive and accurate predictions.

Predicting and Interpreting Protein Developability Via Transfer of Convolutional Sequence Representation

Engineered proteins have emerged as novel diagnostics, therapeutics, and catalysts. Often, poor protein developability─quantified by expression, solubility, and stability─hinders utility. The ability to predict protein developability from amino acid sequence would reduce the experimental burden when selecting candidates. Recent advances in screening technologies enabled a high-throughput (HT) developability dataset for 105 of 1020 possible variants of protein ligand scaffold Gp2. In this work, we evaluate the ability of neural networks to learn a developability representation from a HT dataset and transfer this knowledge to predict recombinant expression beyond observed sequences. The model convolves learned amino acid properties to predict expression levels 44% closer to the experimental variance compared to a non-embedded control. Analysis of learned amino acid embeddings highlights the uniqueness of cysteine, the importance of hydrophobicity and charge, and the unimportance of aromaticity, when aiming to improve the developability of small proteins. We identify clusters of similar sequences with increased recombinant expression through nonlinear dimensionality reduction and we explore the inferred expression landscape via nested sampling. The analysis enables the first direct visualization of the fitness landscape and highlights the existence of evolutionary bottlenecks in sequence space giving rise to competing subpopulations of sequences with different developability. The work advances applied protein engineering efforts by predicting and interpreting protein scaffold expression from a limited dataset. Furthermore, our statistical mechanical treatment of the problem advances foundational efforts to characterize the structure of the protein fitness landscape and the amino acid characteristics that influence protein developability.

DeCOIL: Optimization of Degenerate Codon Libraries for Machine Learning-Assisted Protein Engineering

With advances in machine learning (ML)-assisted protein engineering, models based on data, biophysics, and natural evolution are being used to propose informed libraries of protein variants to explore. Synthesizing these libraries for experimental screens is a major bottleneck, as the cost of obtaining large numbers of exact gene sequences is often prohibitive. Degenerate codon (DC) libraries are a cost-effective alternative for generating combinatorial mutagenesis libraries where mutations are targeted to a handful of amino acid sites. However, existing computational methods to optimize DC libraries to include desired protein variants are not well suited to design libraries for ML-assisted protein engineering. To address these drawbacks, we present DEgenerate Codon Optimization for Informed Libraries (DeCOIL), a generalized method that directly optimizes DC libraries to be useful for protein engineering: to sample protein variants that are likely to have both high fitness and high diversity in the sequence search space. Using computational simulations and wet-lab experiments, we demonstrate that DeCOIL is effective across two specific case studies, with the potential to be applied to many other use cases. DeCOIL offers several advantages over existing methods, as it is direct, easy to use, generalizable, and scalable. With accompanying software (https://github.com/jsunn-y/DeCOIL), DeCOIL can be readily implemented to generate desired informed libraries.

In vitro continuous protein evolution empowered by machine learning and automation

Directed evolution has become one of the most successful and powerful tools for protein engineering. However, the efforts required for designing, constructing, and screening a large library of variants can be laborious, time-consuming, and costly. With the recent advent of machine learning (ML) in the directed evolution of proteins, researchers can now evaluate variants in silico and guide a more efficient directed evolution campaign. Furthermore, recent advancements in laboratory automation have enabled the rapid execution of long, complex experiments for high-throughput data acquisition in both industrial and academic settings, thus providing the means to collect a large quantity of data required to develop ML models for protein engineering. In this perspective, we propose a closed-loop in vitro continuous protein evolution framework that leverages the best of both worlds, ML and automation, and provide a brief overview of the recent developments in the field.

Simplifying complex antibody engineering using machine learning

Machine learning is transforming antibody engineering by enabling the generation of drug-like monoclonal antibodies with unprecedented efficiency. Unsupervised algorithms trained on massive and diverse protein sequence datasets facilitate the prediction of panels of antibody variants with native-like intrinsic properties (e.g., high stability), greatly reducing the amount of subsequent experimentation needed to identify specific candidates that also possess desired extrinsic properties (e.g., high affinity). Additionally, supervised algorithms, which are trained on deep sequencing datasets obtained after enrichment of in vitro antibody libraries for one or more specific extrinsic properties, enable the prediction of antibody variants with desired combinations of extrinsic properties without the need for additional screening. Here we review recent advances using both machine learning approaches and how they are impacting the field of antibody engineering as well as key outstanding challenges and opportunities for these paradigm-changing methods.

Learning protein fitness landscapes with deep mutational scanning data from multiple sources

One of the key points of machine learning-assisted directed evolution (MLDE) is the accurate learning of the fitness landscape, a conceptual mapping from sequence variants to the desired function. Here, we describe a multi-protein training scheme that leverages the existing deep mutational scanning data from diverse proteins to aid in understanding the fitness landscape of a new protein. Proof-of-concept trials are designed to validate this training scheme in three aspects: random and positional extrapolation for single-variant effects, zero-shot fitness predictions for new proteins, and extrapolation for higher-order variant effects from single-variant effects. Moreover, our study identified previously overlooked strong baselines, and their unexpectedly good performance brings our attention to the pitfalls of MLDE. Overall, these results may improve our understanding of the association between different protein fitness profiles and shed light on developing better machine learning-assisted approaches to the directed evolution of proteins. A record of this paper's transparent peer review process is included in the supplemental information.

OpenProteinSet: Training data for structural biology at scale

Multiple sequence alignments (MSAs) of proteins encode rich biological information and have been workhorses in bioinformatic methods for tasks like protein design and protein structure prediction for decades. Recent breakthroughs like AlphaFold2 that use transformers to attend directly over large quantities of raw MSAs have reaffirmed their importance. Generation of MSAs is highly computationally intensive, however, and no datasets comparable to those used to train AlphaFold2 have been made available to the research community, hindering progress in machine learning for proteins. To remedy this problem, we introduce OpenProteinSet, an open-source corpus of more than 16 million MSAs, associated structural homologs from the Protein Data Bank, and AlphaFold2 protein structure predictions. We have previously demonstrated the utility of OpenProteinSet by successfully retraining AlphaFold2 on it. We expect OpenProteinSet to be broadly useful as training and validation data for 1) diverse tasks focused on protein structure, function, and design and 2) large-scale multimodal machine learning research.

Updated benchmarking of variant effect predictors using deep mutational scanning

The assessment of variant effect predictor (VEP) performance is fraught with biases introduced by benchmarking against clinical observations. In this study, building on our previous work, we use independently generated measurements of protein function from deep mutational scanning (DMS) experiments for 26 human proteins to benchmark 55 different VEPs, while introducing minimal data circularity. Many top-performing VEPs are unsupervised methods including EVE, DeepSequence and ESM-1v, a protein language model that ranked first overall. However, the strong performance of recent supervised VEPs, in particular VARITY, shows that developers are taking data circularity and bias issues seriously. We also assess the performance of DMS and unsupervised VEPs for discriminating between known pathogenic and putatively benign missense variants. Our findings are mixed, demonstrating that some DMS datasets perform exceptionally at variant classification, while others are poor. Notably, we observe a striking correlation between VEP agreement with DMS data and performance in identifying clinically relevant variants, strongly supporting the validity of our rankings and the utility of DMS for independent benchmarking.

Cross-protein transfer learning substantially improves disease variant prediction

BACKGROUND

Genetic variation in the human genome is a major determinant of individual disease risk, but the vast majority of missense variants have unknown etiological effects. Here, we present a robust learning framework for leveraging saturation mutagenesis experiments to construct accurate computational predictors of proteome-wide missense variant pathogenicity.

RESULTS

We train cross-protein transfer (CPT) models using deep mutational scanning (DMS) data from only five proteins and achieve state-of-the-art performance on clinical variant interpretation for unseen proteins across the human proteome. We also improve predictive accuracy on DMS data from held-out proteins. High sensitivity is crucial for clinical applications and our model CPT-1 particularly excels in this regime. For instance, at 95% sensitivity of detecting human disease variants annotated in ClinVar, CPT-1 improves specificity to 68%, from 27% for ESM-1v and 55% for EVE. Furthermore, for genes not used to train REVEL, a supervised method widely used by clinicians, we show that CPT-1 compares favorably with REVEL. Our framework combines predictive features derived from general protein sequence models, vertebrate sequence alignments, and AlphaFold structures, and it is adaptable to the future inclusion of other sources of information. We find that vertebrate alignments, albeit rather shallow with only 100 genomes, provide a strong signal for variant pathogenicity prediction that is complementary to recent deep learning-based models trained on massive amounts of protein sequence data. We release predictions for all possible missense variants in 90% of human genes.

CONCLUSIONS

Our results demonstrate the utility of mutational scanning data for learning properties of variants that transfer to unseen proteins.

Structure-Informed Protein Language Models are Robust Predictors for Variant Effects

Predicting protein variant effects through machine learning is often challenged by the scarcity of experimentally measured effect labels. Recently, protein language models (pLMs) emerge as zero-shot predictors without the need of effect labels, by modeling the evolutionary distribution of functional protein sequences. However, biological contexts important to variant effects are implicitly modeled and effectively marginalized. By assessing the sequence awareness and the structure awareness of pLMs, we find that their improvements often correlate with better variant effect prediction but their tradeoff can present a barrier as observed in over-finetuning to specific family sequences. We introduce a framework of structure-informed pLMs (SI-pLMs) to inject protein structural contexts purposely and controllably, by extending masked sequence denoising in conventional pLMs to cross-modality denoising. Our SI-pLMs are applicable to revising any sequence-only pLMs through model architecture and training objectives. They do not require structure data as model inputs for variant effect prediction and only use structures as context provider and model regularizer during training. Numerical results over deep mutagenesis scanning benchmarks show that our SI-pLMs, despite relatively compact sizes, are robustly top performers against competing methods including other pLMs, regardless of the target protein family's evolutionary information content or the tendency to overfitting / over-finetuning. Learned distributions in structural contexts could enhance sequence distributions in predicting variant effects. Ablation studies reveal major contributing factors and analyses of sequence embeddings provide further insights. The data and scripts are available at https://github.com/Stephen2526/Structure-informed_PLM.git.

Mega-scale experimental analysis of protein folding stability in biology and design

Advances in DNA sequencing and machine learning are providing insights into protein sequences and structures on an enormous scale1. However, the energetics driving folding are invisible in these structures and remain largely unknown2. The hidden thermodynamics of folding can drive disease3,4, shape protein evolution5-7 and guide protein engineering8-10, and new approaches are needed to reveal these thermodynamics for every sequence and structure. Here we present cDNA display proteolysis, a method for measuring thermodynamic folding stability for up to 900,000 protein domains in a one-week experiment. From 1.8 million measurements in total, we curated a set of around 776,000 high-quality folding stabilities covering all single amino acid variants and selected double mutants of 331 natural and 148 de novo designed protein domains 40-72 amino acids in length. Using this extensive dataset, we quantified (1) environmental factors influencing amino acid fitness, (2) thermodynamic couplings (including unexpected interactions) between protein sites, and (3) the global divergence between evolutionary amino acid usage and protein folding stability. We also examined how our approach could identify stability determinants in designed proteins and evaluate design methods. The cDNA display proteolysis method is fast, accurate and uniquely scalable, and promises to reveal the quantitative rules for how amino acid sequences encode folding stability.

Artificial intelligence-aided protein engineering: from topological data analysis to deep protein language models

Protein engineering is an emerging field in biotechnology that has the potential to revolutionize various areas, such as antibody design, drug discovery, food security, ecology, and more. However, the mutational space involved is too vast to be handled through experimental means alone. Leveraging accumulative protein databases, machine learning (ML) models, particularly those based on natural language processing (NLP), have considerably expedited protein engineering. Moreover, advances in topological data analysis (TDA) and artificial intelligence-based protein structure prediction, such as AlphaFold2, have made more powerful structure-based ML-assisted protein engineering strategies possible. This review aims to offer a comprehensive, systematic, and indispensable set of methodological components, including TDA and NLP, for protein engineering and to facilitate their future development.

EvoRator2: Predicting Site-specific Amino Acid Substitutions Based on Protein Structural Information Using Deep Learning

Multiple sequence alignments (MSAs) are the workhorse of molecular evolution and structural biology research. From MSAs, the amino acids that are tolerated at each site during protein evolution can be inferred. However, little is known regarding the repertoire of tolerated amino acids in proteins when only a few or no sequence homologs are available, such as orphan and de novo designed proteins. Here we present EvoRator2, a deep-learning algorithm trained on over 15,000 protein structures that can predict which amino acids are tolerated at any given site, based exclusively on protein structural information mined from atomic coordinate files. We show that EvoRator2 obtained satisfying results for the prediction of position-weighted scoring matrices (PSSM). We further show that EvoRator2 obtained near state-of-the-art performance on proteins with high quality structures in predicting the effect of mutations in deep mutation scanning (DMS) experiments and that for certain DMS targets, EvoRator2 outperformed state-of-the-art methods. We also show that by combining EvoRator2's predictions with those obtained by a state-of-the-art deep-learning method that accounts for the information in the MSA, the prediction of the effect of mutation in DMS experiments was improved in terms of both accuracy and stability. EvoRator2 is designed to predict which amino-acid substitutions are tolerated in such proteins without many homologous sequences, including orphan or de novo designed proteins. We implemented our approach in the EvoRator web server (https://evorator.tau.ac.il).

Discovering functionally important sites in proteins

Proteins play important roles in biology, biotechnology and pharmacology, and missense variants are a common cause of disease. Discovering functionally important sites in proteins is a central but difficult problem because of the lack of large, systematic data sets. Sequence conservation can highlight residues that are functionally important but is often convoluted with a signal for preserving structural stability. We here present a machine learning method to predict functional sites by combining statistical models for protein sequences with biophysical models of stability. We train the model using multiplexed experimental data on variant effects and validate it broadly. We show how the model can be used to discover active sites, as well as regulatory and binding sites. We illustrate the utility of the model by prospective prediction and subsequent experimental validation on the functional consequences of missense variants in HPRT1 which may cause Lesch-Nyhan syndrome, and pinpoint the molecular mechanisms by which they cause disease.

From genotype to phenotype: computational approaches for inferring microbial traits relevant to the food industry

When selecting microbial strains for the production of fermented foods, various microbial phenotypes need to be taken into account to achieve target product characteristics, such as biosafety, flavor, texture, and health-promoting effects. Through continuous advances in sequencing technologies, microbial whole-genome sequences of increasing quality can now be obtained both cheaper and faster, which increases the relevance of genome-based characterization of microbial phenotypes. Prediction of microbial phenotypes from genome sequences makes it possible to quickly screen large strain collections in silico to identify candidates with desirable traits. Several microbial phenotypes relevant to the production of fermented foods can be predicted using knowledge-based approaches, leveraging our existing understanding of the genetic and molecular mechanisms underlying those phenotypes. In the absence of this knowledge, data-driven approaches can be applied to estimate genotype-phenotype relationships based on large experimental datasets. Here, we review computational methods that implement knowledge- and data-driven approaches for phenotype prediction, as well as methods that combine elements from both approaches. Furthermore, we provide examples of how these methods have been applied in industrial biotechnology, with special focus on the fermented food industry.

An Atlas of Variant Effects to understand the genome at nucleotide resolution

Sequencing has revealed hundreds of millions of human genetic variants, and continued efforts will only add to this variant avalanche. Insufficient information exists to interpret the effects of most variants, limiting opportunities for precision medicine and comprehension of genome function. A solution lies in experimental assessment of the functional effect of variants, which can reveal their biological and clinical impact. However, variant effect assays have generally been undertaken reactively for individual variants only after and, in most cases long after, their first observation. Now, multiplexed assays of variant effect can characterise massive numbers of variants simultaneously, yielding variant effect maps that reveal the function of every possible single nucleotide change in a gene or regulatory element. Generating maps for every protein encoding gene and regulatory element in the human genome would create an 'Atlas' of variant effect maps and transform our understanding of genetics and usher in a new era of nucleotide-resolution functional knowledge of the genome. An Atlas would reveal the fundamental biology of the human genome, inform human evolution, empower the development and use of therapeutics and maximize the utility of genomics for diagnosing and treating disease. The Atlas of Variant Effects Alliance is an international collaborative group comprising hundreds of researchers, technologists and clinicians dedicated to realising an Atlas of Variant Effects to help deliver on the promise of genomics.

Correspondence between functional scores from deep mutational scans and predicted effects on protein stability

Many methodologically diverse computational methods have been applied to the growing challenge of predicting and interpreting the effects of protein variants. As many pathogenic mutations have a perturbing effect on protein stability or intermolecular interactions, one highly interpretable approach is to use protein structural information to model the physical impacts of variants and predict their likely effects on protein stability and interactions. Previous efforts have assessed the accuracy of stability predictors in reproducing thermodynamically accurate values and evaluated their ability to distinguish between known pathogenic and benign mutations. Here, we take an alternate approach, and explore how well stability predictor scores correlate with functional impacts derived from deep mutational scanning (DMS) experiments. In this work, we compare the predictions of 9 protein stability-based tools against mutant protein fitness values from 49 independent DMS datasets, covering 170,940 unique single amino acid variants. We find that FoldX and Rosetta show the strongest correlations with DMS-based functional scores, similar to their previous top performance in distinguishing between pathogenic and benign variants. For both methods, performance is considerably improved when considering intermolecular interactions from protein complex structures, when available. Furthermore, using these two predictors, we derive a "Foldetta" consensus score, which improves upon the performance of both, and manages to match dedicated variant effect predictors in reflecting variant functional impacts. Finally, we also highlight that predicted stability effects show consistently higher correlations with certain DMS experimental phenotypes, particularly those based upon protein abundance, and, in certain cases, can significantly outcompete sequence-based variant effect prediction methodologies for predicting functional scores from DMS experiments.

Predicting the antigenic evolution of SARS-COV-2 with deep learning

The relentless evolution of SARS-CoV-2 poses a significant threat to public health, as it adapts to immune pressure from vaccines and natural infections. Gaining insights into potential antigenic changes is critical but challenging due to the vast sequence space. Here, we introduce the Machine Learning-guided Antigenic Evolution Prediction (MLAEP), which combines structure modeling, multi-task learning, and genetic algorithms to predict the viral fitness landscape and explore antigenic evolution via in silico directed evolution. By analyzing existing SARS-CoV-2 variants, MLAEP accurately infers variant order along antigenic evolutionary trajectories, correlating with corresponding sampling time. Our approach identified novel mutations in immunocompromised COVID-19 patients and emerging variants like XBB1.5. Additionally, MLAEP predictions were validated through in vitro neutralizing antibody binding assays, demonstrating that the predicted variants exhibited enhanced immune evasion. By profiling existing variants and predicting potential antigenic changes, MLAEP aids in vaccine development and enhances preparedness against future SARS-CoV-2 variants.

Structural and computational design of a SARS-CoV-2 spike antigen with improved expression and immunogenicity

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) variants of concern challenge the efficacy of approved vaccines, emphasizing the need for updated spike antigens. Here, we use an evolutionary-based design aimed at boosting protein expression levels of S-2P and improving immunogenic outcomes in mice. Thirty-six prototype antigens were generated in silico and 15 were produced for biochemical analysis. S2D14, which contains 20 computationally designed mutations within the S2 domain and a rationally engineered D614G mutation in the SD2 domain, has an ~11-fold increase in protein yield and retains RBD antigenicity. Cryo-electron microscopy structures reveal a mixture of populations in various RBD conformational states. Vaccination of mice with adjuvanted S2D14 elicited higher cross-neutralizing antibody titers than adjuvanted S-2P against the SARS-CoV-2 Wuhan strain and four variants of concern. S2D14 may be a useful scaffold or tool for the design of future coronavirus vaccines, and the approaches used for the design of S2D14 may be broadly applicable to streamline vaccine discovery.

The landscape of tolerated genetic variation in humans and primates

Personalized genome sequencing has revealed millions of genetic differences between individuals, but our understanding of their clinical relevance remains largely incomplete. To systematically decipher the effects of human genetic variants, we obtained whole-genome sequencing data for 809 individuals from 233 primate species and identified 4.3 million common protein-altering variants with orthologs in humans. We show that these variants can be inferred to have nondeleterious effects in humans based on their presence at high allele frequencies in other primate populations. We use this resource to classify 6% of all possible human protein-altering variants as likely benign and impute the pathogenicity of the remaining 94% of variants with deep learning, achieving state-of-the-art accuracy for diagnosing pathogenic variants in patients with genetic diseases.

GUD-VE visualization tool for physicochemical properties of proteins

The physicochemical properties of primary sequences of proteins helps in determining both the structure and biological functions. The sequence analysis of the proteins and nucleic acids is most fundamental element of bioinformatics. Without these elements, it is impossible to gain insight deeper molecular and biochemical mechanisms. For this purpose, the computational methods like bioinformatics tools assist experts and novices alike in resolving issues relating to protein analysis. Similarly, this proposed work, for the graphical user interface (GUI) based prediction and visualization through the computations-based method done on Jupyter Notebook with tkinter package which allows the creation of a program on a local host platform and accessed by the programmer.•When it is queried with a protein sequence, it predicts physicochemical parameters of the peptides.•Users can choose to visualize the findings acquired either anonymously or on the user-specified email address and compare the biophysical properties of one protein with other using amino acids (AA) sequences. The aim of this paper is to meet the requirements of experimentalists, not just hardcore bioinformaticians related to biophysical properties prediction and comparison with other proteins. The code for it has been uploaded on GitHub (an online repository of codes) in private mode.

Machine Learning for Protein Engineering

Directed evolution of proteins has been the most effective method for protein engineering. However, a new paradigm is emerging, fusing the library generation and screening approaches of traditional directed evolution with computation through the training of machine learning models on protein sequence fitness data. This chapter highlights successful applications of machine learning to protein engineering and directed evolution, organized by the improvements that have been made with respect to each step of the directed evolution cycle. Additionally, we provide an outlook for the future based on the current direction of the field, namely in the development of calibrated models and in incorporating other modalities, such as protein structure.

Unsupervised Deep Learning Can Identify Protein Functional Groups from Unaligned Sequences

Interpreting protein function from sequence data is a fundamental goal of bioinformatics. However, our current understanding of protein diversity is bottlenecked by the fact that most proteins have only been functionally validated in model organisms, limiting our understanding of how function varies with gene sequence diversity. Thus, accuracy of inferences in clades without model representatives is questionable. Unsupervised learning may help to ameliorate this bias by identifying highly complex patterns and structure from large datasets without external labels. Here we present DeepSeqProt, an unsupervised deep learning program for exploring large protein sequence datasets. DeepSeqProt is a clustering tool capable of distinguishing between broad classes of proteins while learning local and global structure of functional space. DeepSeqProt is capable of learning salient biological features from unaligned, unannotated sequences. DeepSeqProt is more likely to capture complete protein families and statistically significant shared ontologies within proteomes than other clustering methods. We hope this framework will prove of use to researchers and provide a preliminary step in further developing unsupervised deep learning in molecular biology.

Rapid protein stability prediction using deep learning representations

Predicting the thermodynamic stability of proteins is a common and widely used step in protein engineering, and when elucidating the molecular mechanisms behind evolution and disease. Here, we present RaSP, a method for making rapid and accurate predictions of changes in protein stability by leveraging deep learning representations. RaSP performs on-par with biophysics-based methods and enables saturation mutagenesis stability predictions in less than a second per residue. We use RaSP to calculate ∼ 230 million stability changes for nearly all single amino acid changes in the human proteome, and examine variants observed in the human population. We find that variants that are common in the population are substantially depleted for severe destabilization, and that there are substantial differences between benign and pathogenic variants, highlighting the role of protein stability in genetic diseases. RaSP is freely available-including via a Web interface-and enables large-scale analyses of stability in experimental and predicted protein structures.

Simultaneous enhancement of multiple functional properties using evolution-informed protein design

Designing optimized proteins is important for a range of practical applications. Protein design is a rapidly developing field that would benefit from approaches that enable many changes in the amino acid primary sequence, rather than a small number of mutations, while maintaining structure and enhancing function. Homologous protein sequences contain extensive information about various protein properties and activities that have emerged over billions of years of evolution. Evolutionary models of sequence co-variation, derived from a set of homologous sequences, have proven effective in a range of applications including structure determination and mutation effect prediction. In this work we apply one of these models (EVcouplings) to computationally design highly divergent variants of the model protein TEM-1 β-lactamase, and characterize these designs experimentally using multiple biochemical and biophysical assays. Nearly all designed variants were functional, including one with 84 mutations from the nearest natural homolog. Surprisingly, all functional designs had large increases in thermostability and most had a broadening of available substrates. These property enhancements occurred while maintaining a nearly identical structure to the wild type enzyme. Collectively, this work demonstrates that evolutionary models of sequence co-variation (1) are able to capture complex epistatic interactions that successfully guide large sequence departures from natural contexts, and (2) can be applied to generate functional diversity useful for many applications in protein design.

High-throughput deep learning variant effect prediction with Sequence UNET

Understanding coding mutations is important for many applications in biology and medicine but the vast mutation space makes comprehensive experimental characterisation impossible. Current predictors are often computationally intensive and difficult to scale, including recent deep learning models. We introduce Sequence UNET, a highly scalable deep learning architecture that classifies and predicts variant frequency from sequence alone using multi-scale representations from a fully convolutional compression/expansion architecture. It achieves comparable pathogenicity prediction to recent methods. We demonstrate scalability by analysing 8.3B variants in 904,134 proteins detected through large-scale proteomics. Sequence UNET runs on modest hardware with a simple Python package.

The landscape of tolerated genetic variation in humans and primates

Personalized genome sequencing has revealed millions of genetic differences between individuals, but our understanding of their clinical relevance remains largely incomplete. To systematically decipher the effects of human genetic variants, we obtained whole genome sequencing data for 809 individuals from 233 primate species, and identified 4.3 million common protein-altering variants with orthologs in human. We show that these variants can be inferred to have non-deleterious effects in human based on their presence at high allele frequencies in other primate populations. We use this resource to classify 6% of all possible human protein-altering variants as likely benign and impute the pathogenicity of the remaining 94% of variants with deep learning, achieving state-of-the-art accuracy for diagnosing pathogenic variants in patients with genetic diseases.

One Sentence Summary

Deep learning classifier trained on 4.3 million common primate missense variants predicts variant pathogenicity in humans.