Deep mutational scanning of an RRM domain of the Saccharomyces cerevisiae poly(A)-binding protein

Deaminase-based RNA recording enables high throughput mutational profiling of protein-RNA interactions

Protein-RNA interactions govern nearly every aspect of RNA metabolism and are frequently dysregulated in disease. While individual protein residues and RNA nucleotides critical for these interactions have been characterized, scalable methods that jointly map protein- and RNA-level determinants remain limited. RNA deaminase fusions have emerged as a powerful strategy to identify transcriptome-wide targets of RNA-binding proteins by converting binding events into site-specific nucleotide edits. Here, we demonstrate that this 'RNA recording' approach enables high-throughput mutational scanning of protein-RNA interfaces. Using the λN-boxB system as a model, we show that editing by a fused TadA adenosine deaminase directly correlates with binding affinity between protein and RNA variants in vitro . Systematic variation of RNA sequence context reveals a strong bias for editing at UA dinucleotides by the engineered TadA8.20, mirroring wild-type TadA preferences. We further demonstrate that stepwise recruitment of the deaminase using nanobody and protein A/G fusions maintains both sequence and binding specificity. Stable expression of the TadA fusion in human cells reproduces in vitro editing patterns across a library of RNA variants. Finally, comprehensive single amino acid mutagenesis of λN in human cells reveals critical residues mediating RNA binding. Together, our results establish RNA recording as a versatile and scalable tool for dissecting protein-RNA interactions at nucleotide and residue resolution, both in vitro and in cells.

Widespread epistasis shapes RNA Polymerase II active site function and evolution

Multi-subunit RNA Polymerases (msRNAPs) are responsible for transcription in all kingdoms of life. These enzymes rely on dynamic, highly conserved active site domains such as the so-called "trigger loop" (TL) to accomplish steps in the transcription cycle. Mutations in the RNA polymerase II (Pol II) TL confer a spectrum of biochemical and genetic phenotypes that suggest two main classes, which decrease or increase catalysis or other nucleotide addition cycle (NAC) events. The Pol II active site relies on networks of residue interactions to function and mutations likely perturb these networks in ways that may alter mechanisms. We have undertaken a structural genetics approach to reveal residue interactions within and surrounding the Pol II TL - determining its "interaction landscape" - by deep mutational scanning in Saccharomyces cerevisiae Pol II. This analysis reveals connections between TL residues and surrounding domains, demonstrating that TL function is tightly coupled to its specific enzyme context.

MMRT: MultiMut Recursive Tree for predicting functional effects of high-order protein variants from low-order variants

Protein sequences primarily determine their stability and functions. Mutations may occur at one, two, or three positions at the same time (low-order variants) or at multiple positions simultaneously (high-order variants), which affect protein functions. So far, low-order variants, such as single variants, double variants, and triple variants, have been well-studied through high-throughput experimental scanning techniques and computational prediction methods. However, research on high-order variants remains limited because of the difficulty of scanning an exponentially large number of potential variant combinations. Nonetheless, studying higher-order variants is crucial for understanding the pathogenesis of complex diseases, advancing protein engineering, and driving precision medicine. In this work, we introduce a novel deep learning model, namely MultiMut Recursive Tree (MMRT), to address this challenge of predicting the functional effects of high-order variants. MMRT integrates deep learning with a recursive tree framework to leverage the information from low-order variants to predict functional effects of high-order variants. We evaluated MMRT on datasets comprising 685,593 high-order variants. Our results (mean Spearman's correlation coefficient 0.55) demonstrated that MMRT outperformed three existing state-of-the-art methods: ESM (evolutionary scale modeling), DeepSequence, and ECNet (evolutionary context-integrated neural network). MMRT thus provides more accurate prediction of the functional effects of high-order protein variants, offering great potential for aiding the interpretation of variants in human disease studies.

Deep mutational scanning of the Trypanosoma brucei developmental regulator RBP6 reveals an essential disordered region influenced by positive residues

To regain infectivity, Trypanosoma brucei, the pathogen causing Human and Animal African trypanosomiasis, undergoes a complex developmental program within the tsetse fly known as metacyclogenesis. RNA-binding protein 6 (RBP6) is a potent orchestrator of this process, however, an understanding of its functionally important domains and their mutational constraints is lacking. Here, we perform deep mutational scanning of the entire RBP6 primary structure. Expression of libraries containing all single-point variants of RBP6 in non-infectious procyclic forms and subsequent purification of infectious metacyclics supports the existence of an RNA-recognition motif (RRM) and reveal an N-terminal intrinsically disordered region (N-IDR). In contrast to the RRM, the N-IDR is more tolerant to substitutions; however, a handful of positions contain a third of all deleterious mutations found in the N-IDR. Introduction of positively charged residues in the N-IDR dramatically alters the normal metacyclogenesis pattern. Our results reveal an essential N-IDR, possibly playing a regulatory role, and an RRM likely involved in protein-RNA interactions.

Biophysics-based protein language models for protein engineering

Protein language models trained on evolutionary data have emerged as powerful tools for predictive problems involving protein sequence, structure, and function. However, these models overlook decades of research into biophysical factors governing protein function. We propose Mutational Effect Transfer Learning (METL), a protein language model framework that unites advanced machine learning and biophysical modeling. Using the METL framework, we pretrain transformer-based neural networks on biophysical simulation data to capture fundamental relationships between protein sequence, structure, and energetics. We finetune METL on experimental sequence-function data to harness these biophysical signals and apply them when predicting protein properties like thermostability, catalytic activity, and fluorescence. METL excels in challenging protein engineering tasks like generalizing from small training sets and position extrapolation, although existing methods that train on evolutionary signals remain powerful for many types of experimental assays. We demonstrate METL's ability to design functional green fluorescent protein variants when trained on only 64 examples, showcasing the potential of biophysics-based protein language models for protein engineering.

Tailoring industrial enzymes for thermostability and activity evolution by the machine learning-based iCASE strategy

The pursuit of obtaining enzymes with high activity and stability remains a grail in enzyme evolution due to the stability-activity trade-off. Here, we develop an isothermal compressibility-assisted dynamic squeezing index perturbation engineering (iCASE) strategy to construct hierarchical modular networks for enzymes of varying complexity. Molecular mechanism analysis elucidates that the peak of adaptive evolution is reached through a structural response mechanism among variants. Furthermore, this dynamic response predictive model using structure-based supervised machine learning is established to predict enzyme function and fitness, demonstrating robust performance across different datasets and reliable prediction for epistasis. The universality of the iCASE strategy is validated by four sorts of enzymes with different structures and catalytic types. This machine learning-based iCASE strategy provides guidance for future research on the fitness evolution of enzymes.

DeepPFP: a multi-task-aware architecture for protein function prediction

Deriving protein function from protein sequences poses a significant challenge due to the intricate relationship between sequence and function. Deep learning has made remarkable strides in predicting sequence-function relationships. However, models tailored for specific tasks or protein types encounter difficulties when using transfer learning across domains. This is attributed to the fact that protein function relies heavily on structural characteristics rather than mere sequence information. Consequently, there is a pressing need for a model capable of capturing shared features among diverse sequence-function mapping tasks to address the generalization issue. In this study, we explore the potential of Model-Agnostic Meta-Learning combined with a protein language model called Evolutionary Scale Modeling to tackle this challenge. Our approach involves training the architecture on five out-domain deep mutational scanning (DMS) datasets and evaluating its performance across four key dimensions. Our findings demonstrate that the proposed architecture exhibits satisfactory performance in terms of generalization and employs an effective few-shot learning strategy. To explain further, Compared to the best results, the Pearson's correlation coefficient (PCC) in the final stage increased by ~0.31%. Furthermore, we leverage the trained architecture to predict binding affinity scores of the DMS dataset of SARS-CoV-2 using transfer learning. Notably, training on a subset of the Ube4b dataset with 500 samples resulted in a notable improvement of 0.11 in the PCC. These results underscore the potential of our conceptual architecture as a promising methodology for multi-task protein function prediction.

Higher-order epistasis within Pol II trigger loop haplotypes

RNA polymerase II (Pol II) has a highly conserved domain, the trigger loop (TL), that controls transcription fidelity and speed. We previously probed pairwise genetic interactions between residues within and surrounding the TL for the purpose of understand functional interactions between residues and to understand how individual mutants might alter TL function. We identified widespread incompatibility between TLs of different species when placed in the Saccharomyces cerevisiae Pol II context, indicating species-specific interactions between otherwise highly conserved TLs and its surroundings. These interactions represent epistasis between TL residues and the rest of Pol II. We sought to understand why certain TL sequences are incompatible with S. cerevisiae Pol II and to dissect the nature of genetic interactions within multiply substituted TLs as a window on higher order epistasis in this system. We identified both positive and negative higher-order residue interactions within example TL haplotypes. Intricate higher-order epistasis formed by TL residues was sometimes only apparent from analysis of intermediate genotypes, emphasizing complexity of epistatic interactions. Furthermore, we distinguished TL substitutions with distinct classes of epistatic patterns, suggesting specific TL residues that potentially influence TL evolution. Our examples of complex residue interactions suggest possible pathways for epistasis to facilitate Pol II evolution.

Understanding epistatic networks in the B1 β-lactamases through coevolutionary statistical modeling and deep mutational scanning

Throughout evolution, protein families undergo substantial sequence divergence while preserving structure and function. Although most mutations are deleterious, evolution can explore sequence space via epistatic networks of intramolecular interactions that alleviate the harmful mutations. However, comprehensive analysis of such epistatic networks across protein families remains limited. Thus, we conduct a family wide analysis of the B1 metallo-β-lactamases, combining experiments (deep mutational scanning, DMS) on two distant homologs (NDM-1 and VIM-2) and computational analyses (in silico DMS based on Direct Coupling Analysis, DCA) of 100 homologs. The methods jointly reveal and quantify prevalent epistasis, as ~1/3rd of equivalent mutations are epistatic in DMS. From DCA, half of the positions have a >6.5 fold difference in effective number of tolerated mutations across the entire family. Notably, both methods locate residues with the strongest epistasis in regions of intermediate residue burial, suggesting a balance of residue packing and mutational freedom in forming epistatic networks. We identify entrenched WT residues between NDM-1 and VIM-2 in DMS, which display statistically distinct behaviors in DCA from non-entrenched residues. Entrenched residues are not easily compensated by changes in single nearby interactions, reinforcing existing findings where a complex epistatic network compounds smaller effects from many interacting residues.

Higher-order epistasis within Pol II trigger loop haplotypes

RNA polymerase II (Pol II) has a highly conserved domain, the trigger loop (TL), that controls transcription fidelity and speed. We previously probed pairwise genetic interactions between residues within and surrounding the TL for the purpose of understand functional interactions between residues and to understand how individual mutants might alter TL function. We identified widespread incompatibility between TLs of different species when placed in the Saccharomyces cerevisiae Pol II context, indicating species-specific interactions between otherwise highly conserved TLs and its surroundings. These interactions represent epistasis between TL residues and the rest of Pol II. We sought to understand why certain TL sequences are incompatible with S. cerevisiae Pol II and to dissect the nature of genetic interactions within multiply substituted TLs as a window on higher order epistasis in this system. We identified both positive and negative higher-order residue interactions within example TL haplotypes. Intricate higher-order epistasis formed by TL residues was sometimes only apparent from analysis of intermediate genotypes, emphasizing complexity of epistatic interactions. Furthermore, we distinguished TL substitutions with distinct classes of epistatic patterns, suggesting specific TL residues that potentially influence TL evolution. Our examples of complex residue interactions suggest possible pathways for epistasis to facilitate Pol II evolution.

Evolutionary Probability and Stacked Regressions Enable Data-Driven Protein Engineering with Minimized Experimental Effort

Protein engineering through directed evolution and (semi)rational approaches is routinely applied to optimize protein properties for a broad range of applications in industry and academia. The multitude of possible variants, combined with limited screening throughput, hampers efficient protein engineering. Data-driven strategies have emerged as a powerful tool to model the protein fitness landscape that can be explored in silico, significantly accelerating protein engineering campaigns. However, such methods require a certain amount of data, which often cannot be provided, to generate a reliable model of the fitness landscape. Here, we introduce MERGE, a method that combines direct coupling analysis (DCA) and machine learning (ML). MERGE enables data-driven protein engineering when only limited data are available for training, typically ranging from 50 to 500 labeled sequences. Our method demonstrates remarkable performance in predicting a protein's fitness value and rank based on its sequence across diverse proteins and properties. Notably, MERGE outperforms state-of-the-art methods when only small data sets are available for modeling, requiring fewer computational resources, and proving particularly promising for protein engineers who have access to limited amounts of data.

Scalable approaches for generating, validating and incorporating data from high-throughput functional assays to improve clinical variant classification

As the adoption and scope of genetic testing continue to expand, interpreting the clinical significance of DNA sequence variants at scale remains a formidable challenge, with a high proportion classified as variants of uncertain significance (VUSs). Genetic testing laboratories have historically relied, in part, on functional data from academic literature to support variant classification. High-throughput functional assays or multiplex assays of variant effect (MAVEs), designed to assess the effects of DNA variants on protein stability and function, represent an important and increasingly available source of evidence for variant classification, but their potential is just beginning to be realized in clinical lab settings. Here, we describe a framework for generating, validating and incorporating data from MAVEs into a semi-quantitative variant classification method applied to clinical genetic testing. Using single-cell gene expression measurements, cellular evidence models were built to assess the effects of DNA variation in 44 genes of clinical interest. This framework was also applied to models for an additional 22 genes with previously published MAVE datasets. In total, modeling data was incorporated from 24 genes into our variant classification method. These data contributed evidence for classifying 4043 observed variants in over 57,000 individuals. Genetic testing laboratories are uniquely positioned to generate, analyze, validate, and incorporate evidence from high-throughput functional data and ultimately enable the use of these data to provide definitive clinical variant classifications for more patients.

Functional effects of mutations in proteins can be predicted and interpreted by guided selection of sequence covariation information

Predicting the effects of one or more mutations to the in vivo or in vitro properties of a wild-type protein is a major computational challenge, due to the presence of epistasis, that is, of interactions between amino acids in the sequence. We introduce a computationally efficient procedure to build minimal epistatic models to predict mutational effects by combining evolutionary (homologous sequence) and few mutational-scan data. Mutagenesis measurements guide the selection of links in a sparse graphical model, while the parameters on the nodes and the edges are inferred from sequence data. We show, on 10 mutational scans, that our pipeline exhibits performances comparable to state-of-the-art deep networks trained on many more data, while requiring much less parameters and being hence more interpretable. In particular, the identified interactions adapt to the wild-type protein and to the fitness or biochemical property experimentally measured, mostly focus on key functional sites, and are not necessarily related to structural contacts. Therefore, our method is able to extract information relevant for one mutational experiment from homologous sequence data reflecting the multitude of structural and functional constraints acting on proteins throughout evolution.

Genotype sampling for deep-learning assisted experimental mapping of a combinatorially complete fitness landscape

MOTIVATION

Experimental characterization of fitness landscapes, which map genotypes onto fitness, is important for both evolutionary biology and protein engineering. It faces a fundamental obstacle in the astronomical number of genotypes whose fitness needs to be measured for any one protein. Deep learning may help to predict the fitness of many genotypes from a smaller neural network training sample of genotypes with experimentally measured fitness. Here I use a recently published experimentally mapped fitness landscape of more than 260 000 protein genotypes to ask how such sampling is best performed.

RESULTS

I show that multilayer perceptrons, recurrent neural networks, convolutional networks, and transformers, can explain more than 90% of fitness variance in the data. In addition, 90% of this performance is reached with a training sample comprising merely ≈103 sequences. Generalization to unseen test data is best when training data is sampled randomly and uniformly, or sampled to minimize the number of synonymous sequences. In contrast, sampling to maximize sequence diversity or codon usage bias reduces performance substantially. These observations hold for more than one network architecture. Simple sampling strategies may perform best when training deep learning neural networks to map fitness landscapes from experimental data.

AVAILABILITY AND IMPLEMENTATION

The fitness landscape data analyzed here is publicly available as described previously (Papkou et al. 2023). All code used to analyze this landscape is publicly available at https://github.com/andreas-wagner-uzh/fitness_landscape_sampling.

Network of epistatic interactions in an enzyme active site revealed by large-scale deep mutational scanning

Cooperative interactions between amino acids are critical for protein function. A genetic reflection of cooperativity is epistasis, which is when a change in the amino acid at one position changes the sequence requirements at another position. To assess epistasis within an enzyme active site, we utilized CTX-M β-lactamase as a model system. CTX-M hydrolyzes β-lactam antibiotics to provide antibiotic resistance, allowing a simple functional selection for rapid sorting of modified enzymes. We created all pairwise mutations across 17 active site positions in the β-lactamase enzyme and quantitated the function of variants against two β-lactam antibiotics using next-generation sequencing. Context-dependent sequence requirements were determined by comparing the antibiotic resistance function of double mutations across the CTX-M active site to their predicted function based on the constituent single mutations, revealing both positive epistasis (synergistic interactions) and negative epistasis (antagonistic interactions) between amino acid substitutions. The resulting trends demonstrate that positive epistasis is present throughout the active site, that epistasis between residues is mediated through substrate interactions, and that residues more tolerant to substitutions serve as generic compensators which are responsible for many cases of positive epistasis. Additionally, we show that a key catalytic residue (Glu166) is amenable to compensatory mutations, and we characterize one such double mutant (E166Y/N170G) that acts by an altered catalytic mechanism. These findings shed light on the unique biochemical factors that drive epistasis within an enzyme active site and will inform enzyme engineering efforts by bridging the gap between amino acid sequence and catalytic function.

Evaluating generalizability of artificial intelligence models for molecular datasets

Deep learning has made rapid advances in modeling molecular sequencing data. Despite achieving high performance on benchmarks, it remains unclear to what extent deep learning models learn general principles and generalize to previously unseen sequences. Benchmarks traditionally interrogate model generalizability by generating metadata based (MB) or sequence-similarity based (SB) train and test splits of input data before assessing model performance. Here, we show that this approach mischaracterizes model generalizability by failing to consider the full spectrum of cross-split overlap, i.e., similarity between train and test splits. We introduce Spectra, a spectral framework for comprehensive model evaluation. For a given model and input data, Spectra plots model performance as a function of decreasing cross-split overlap and reports the area under this curve as a measure of generalizability. We apply Spectra to 18 sequencing datasets with associated phenotypes ranging from antibiotic resistance in tuberculosis to protein-ligand binding to evaluate the generalizability of 19 state-of-the-art deep learning models, including large language models, graph neural networks, diffusion models, and convolutional neural networks. We show that SB and MB splits provide an incomplete assessment of model generalizability. With Spectra, we find as cross-split overlap decreases, deep learning models consistently exhibit a reduction in performance in a task- and model-dependent manner. Although no model consistently achieved the highest performance across all tasks, we show that deep learning models can generalize to previously unseen sequences on specific tasks. Spectra paves the way toward a better understanding of how foundation models generalize in biology.

Protein design using structure-based residue preferences

Recent developments in protein design rely on large neural networks with up to 100s of millions of parameters, yet it is unclear which residue dependencies are critical for determining protein function. Here, we show that amino acid preferences at individual residues-without accounting for mutation interactions-explain much and sometimes virtually all of the combinatorial mutation effects across 8 datasets (R2 ~ 78-98%). Hence, few observations (~100 times the number of mutated residues) enable accurate prediction of held-out variant effects (Pearson r > 0.80). We hypothesized that the local structural contexts around a residue could be sufficient to predict mutation preferences, and develop an unsupervised approach termed CoVES (Combinatorial Variant Effects from Structure). Our results suggest that CoVES outperforms not just model-free methods but also similarly to complex models for creating functional and diverse protein variants. CoVES offers an effective alternative to complicated models for identifying functional protein mutations.

Accurate top protein variant discovery via low-N pick-and-validate machine learning

A strategy to obtain the greatest number of best-performing variants with least amount of experimental effort over the vast combinatorial mutational landscape would have enormous utility in boosting resource producibility for protein engineering. Toward this goal, we present a simple and effective machine learning-based strategy that outperforms other state-of-the-art methods. Our strategy integrates zero-shot prediction and multi-round sampling to direct active learning via experimenting with only a few predicted top variants. We find that four rounds of low-N pick-and-validate sampling of 12 variants for machine learning yielded the best accuracy of up to 92.6% in selecting the true top 1% variants in combinatorial mutant libraries, whereas two rounds of 24 variants can also be used. We demonstrate our strategy in successfully discovering high-performance protein variants from diverse families including the CRISPR-based genome editors, supporting its generalizable application for solving protein engineering tasks. A record of this paper's transparent peer review process is included in the supplemental information.

Positive epistasis drives clavulanic acid resistance in double mutant libraries of BlaC β-lactamase

Phenotypic effects of mutations are highly dependent on the genetic backgrounds in which they occur, due to epistatic effects. To test how easily the loss of enzyme activity can be compensated for, we screen mutant libraries of BlaC, a β-lactamase from Mycobacterium tuberculosis, for fitness in the presence of carbenicillin and the inhibitor clavulanic acid. Using a semi-rational approach and deep sequencing, we prepare four double-site saturation libraries and determine the relative fitness effect for 1534/1540 (99.6%) of the unique library members at two temperatures. Each library comprises variants of a residue known to be relevant for clavulanic acid resistance as well as residue 105, which regulates access to the active site. Variants with greatly improved fitness were identified within each library, demonstrating that compensatory mutations for loss of activity can be readily found. In most cases, the fittest variants are a result of positive epistasis, indicating strong synergistic effects between the chosen residue pairs. Our study sheds light on a role of epistasis in the evolution of functional residues and underlines the highly adaptive potential of BlaC.

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.

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.

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.

CovET: A covariation-evolutionary trace method that identifies protein structure-function modules

Measuring the relative effect that any two sequence positions have on each other may improve protein design or help better interpret coding variants. Current approaches use statistics and machine learning but rarely consider phylogenetic divergences which, as shown by Evolutionary Trace studies, provide insight into the functional impact of sequence perturbations. Here, we reframe covariation analyses in the Evolutionary Trace framework to measure the relative tolerance to perturbation of each residue pair during evolution. This approach (CovET) systematically accounts for phylogenetic divergences: at each divergence event, we penalize covariation patterns that belie evolutionary coupling. We find that while CovET approximates the performance of existing methods to predict individual structural contacts, it performs significantly better at finding structural clusters of coupled residues and ligand binding sites. For example, CovET found more functionally critical residues when we examined the RNA recognition motif and WW domains. It correlates better with large-scale epistasis screen data. In the dopamine D2 receptor, top CovET residue pairs recovered accurately the allosteric activation pathway characterized for Class A G protein-coupled receptors. These data suggest that CovET ranks highest the sequence position pairs that play critical functional roles through epistatic and allosteric interactions in evolutionarily relevant structure-function motifs. CovET complements current methods and may shed light on fundamental molecular mechanisms of protein structure and function.

Evolvability-enhancing mutations in the fitness landscapes of an RNA and a protein

Can evolvability-the ability to produce adaptive heritable variation-itself evolve through adaptive Darwinian evolution? If so, then Darwinian evolution may help create the conditions that enable Darwinian evolution. Here I propose a framework that is suitable to address this question with available experimental data on adaptive landscapes. I introduce the notion of an evolvability-enhancing mutation, which increases the likelihood that subsequent mutations in an evolving organism, protein, or RNA molecule are adaptive. I search for such mutations in the experimentally characterized and combinatorially complete fitness landscapes of a protein and an RNA molecule. I find that such evolvability-enhancing mutations indeed exist. They constitute a small fraction of all mutations, which shift the distribution of fitness effects of subsequent mutations towards less deleterious mutations, and increase the incidence of beneficial mutations. Evolving populations which experience such mutations can evolve significantly higher fitness. The study of evolvability-enhancing mutations opens many avenues of investigation into the evolution of evolvability.

Deep Mutational Scanning of an Oxygen-Independent Fluorescent Protein CreiLOV for Comprehensive Profiling of Mutational and Epistatic Effects

Oxygen-independent, flavin mononucleotide-based fluorescent proteins (FbFPs) are promising alternatives to green fluorescent protein in anaerobic contexts. Deep mutational scanning performs systematic profiling of protein sequence-function relationships but has not been applied to FbFPs. Focusing on CreiLOV from Chlamydomonas reinhardtii, we created and analyzed two comprehensive mutant collections: (1) single-residue, site-saturation mutagenesis libraries covering all 118 residues; and (2) a full combinatorial metagenesis library among 20 mutations at 15 residues, where mutation and residue selection was based on single-site mutagenesis results. Notably, the second type of library is indispensable to study higher-order epistasis but underrepresented in the literature. Using optimized FACS-seq assays, 2,185 (>92.5%) out of 2,360 possible single-site mutants and 165,428 (>89.7%) out of 184,320 possible combinatorial mutants were reliably assigned with fitness values. We constructed statistical and machine-learning models to analyze the CreiLOV data set, enabling accurate fitness prediction of higher-order mutants using lower-order mutagenesis data. In addition, we successfully isolated CreiLOV variants with improved fluorescence quantum yield and thermostability. This work provides new empirical data and design rules to engineer combinatorial protein variants.

Mutational scan inferred binding energetics and structure in intrinsically disordered protein CcdA

Unlike globular proteins, mutational effects on the function of Intrinsically Disordered Proteins (IDPs) are not well-studied. Deep Mutational Scanning of a yeast surface displayed mutant library yields insights into sequence-function relationships in the CcdA IDP. The approach enables facile prediction of interface residues and local structural signatures of the bound conformation. In contrast to previous titration-based approaches which use a number of ligand concentrations, we show that use of a single rationally chosen ligand concentration can provide quantitative estimates of relative binding constants for large numbers of protein variants. This is because the extended interface of IDP ensures that energetic effects of point mutations are spread over a much smaller range than for globular proteins. Our data also provides insights into the much-debated role of helicity and disorder in partner binding of IDPs. Based on this exhaustive mutational sensitivity dataset, a rudimentary model was developed in an attempt to predict mutational effects on binding affinity of IDPs that form alpha-helical structures upon binding.

The Structure of Evolutionary Model Space for Proteins across the Tree of Life

The factors that determine the relative rates of amino acid substitution during protein evolution are complex and known to vary among taxa. We estimated relative exchangeabilities for pairs of amino acids from clades spread across the tree of life and assessed the historical signal in the distances among these clade-specific models. We separately trained these models on collections of arbitrarily selected protein alignments and on ribosomal protein alignments. In both cases, we found a clear separation between the models trained using multiple sequence alignments from bacterial clades and the models trained on archaeal and eukaryotic data. We assessed the predictive power of our novel clade-specific models of sequence evolution by asking whether fit to the models could be used to identify the source of multiple sequence alignments. Model fit was generally able to correctly classify protein alignments at the level of domain (bacterial versus archaeal), but the accuracy of classification at finer scales was much lower. The only exceptions to this were the relatively high classification accuracy for two archaeal lineages: Halobacteriaceae and Thermoprotei. Genomic GC content had a modest impact on relative exchangeabilities despite having a large impact on amino acid frequencies. Relative exchangeabilities involving aromatic residues exhibited the largest differences among models. There were a small number of exchangeabilities that exhibited large differences in comparisons among major clades and between generalized models and ribosomal protein models. Taken as a whole, these results reveal that a small number of relative exchangeabilities are responsible for much of the structure of the "model space" for protein sequence evolution. The clade-specific models we generated may be useful tools for protein phylogenetics, and the structure of evolutionary model space that they revealed has implications for phylogenomic inference across the tree of life.

SESNet: sequence-structure feature-integrated deep learning method for data-efficient protein engineering

Deep learning has been widely used for protein engineering. However, it is limited by the lack of sufficient experimental data to train an accurate model for predicting the functional fitness of high-order mutants. Here, we develop SESNet, a supervised deep-learning model to predict the fitness for protein mutants by leveraging both sequence and structure information, and exploiting attention mechanism. Our model integrates local evolutionary context from homologous sequences, the global evolutionary context encoding rich semantic from the universal protein sequence space and the structure information accounting for the microenvironment around each residue in a protein. We show that SESNet outperforms state-of-the-art models for predicting the sequence-function relationship on 26 deep mutational scanning datasets. More importantly, we propose a data augmentation strategy by leveraging the data from unsupervised models to pre-train our model. After that, our model can achieve strikingly high accuracy in prediction of the fitness of protein mutants, especially for the higher order variants (> 4 mutation sites), when finetuned by using only a small number of experimental mutation data (< 50). The strategy proposed is of great practical value as the required experimental effort, i.e., producing a few tens of experimental mutation data on a given protein, is generally affordable by an ordinary biochemical group and can be applied on almost any protein.

Mutational fitness landscape and drug resistance

Robust technology has been developed to systematically quantify fitness landscapes that provide valuable opportunities to improve our understanding of drug resistance and define new avenues to develop drugs with reduced resistance susceptibility. We outline the critical importance of drug resistance studies and the potential for fitness landscape approaches to contribute to this effort. We describe the major technical advancements in mutational scanning, which is the primary approach used to quantify protein fitness landscapes. There are many complex steps to consider in planning and executing mutational scanning projects including developing a selection scheme, generating mutant libraries, tracking the frequency of variants using next-generation sequencing, and processing and interpreting the data. Key experimental parameters impacting each of these steps are discussed to aid in planning fitness landscape studies. There is a strong need for improved understanding of drug resistance, and fitness landscapes provide a promising new approach.

What Has Genomics Taught An Evolutionary Biologist?

Genomics, an interdisciplinary field of biology on the structure, function, and evolution of genomes, has revolutionized many subdisciplines of life sciences, including my field of evolutionary biology, by supplying huge data, bringing high-throughput technologies, and offering a new approach to biology. In this review, I describe what I have learned from genomics and highlight the fundamental knowledge and mechanistic insights gained. I focus on three broad topics that are central to evolutionary biology and beyond-variation, interaction, and selection-and use primarily my own research and study subjects as examples. In the next decade or two, I expect that the most important contributions of genomics to evolutionary biology will be to provide genome sequences of nearly all known species on Earth, facilitate high-throughput phenotyping of natural variants and systematically constructed mutants for mapping genotype-phenotype-fitness landscapes, and assist the determination of causality in evolutionary processes using experimental evolution.

Integrating deep mutational scanning and low-throughput mutagenesis data to predict the impact of amino acid variants

BACKGROUND

Evaluating the impact of amino acid variants has been a critical challenge for studying protein function and interpreting genomic data. High-throughput experimental methods like deep mutational scanning (DMS) can measure the effect of large numbers of variants in a target protein, but because DMS studies have not been performed on all proteins, researchers also model DMS data computationally to estimate variant impacts by predictors.

RESULTS

In this study, we extended a linear regression-based predictor to explore whether incorporating data from alanine scanning (AS), a widely used low-throughput mutagenesis method, would improve prediction results. To evaluate our model, we collected 146 AS datasets, mapping to 54 DMS datasets across 22 distinct proteins.

CONCLUSIONS

We show that improved model performance depends on the compatibility of the DMS and AS assays, and the scale of improvement is closely related to the correlation between DMS and AS results.

Pab1 acetylation at K131 decreases stress granule formation in Saccharomyces cerevisiae

Under environmental stress, such as glucose deprivation, cells form stress granules-the accumulation of cytoplasmic aggregates of repressed translational initiation complexes, proteins, and stalled mRNAs. Recent research implicates stress granules in various diseases, such as neurodegenerative diseases, but the exact regulators responsible for the assembly and disassembly of stress granules are unknown. An important aspect of stress granule formation is the presence of posttranslational modifications on core proteins. One of those modifications is lysine acetylation, which is regulated by either a lysine acetyltransferase or a lysine deacetylase enzyme. This work deciphers the impact of lysine acetylation on an essential protein found in Saccharomyces cerevisiae stress granules, poly(A)-binding protein (Pab1). We demonstrated that an acetylation mimic of the lysine residue in position 131 reduces stress granule formation upon glucose deprivation and other stressors such as ethanol, raffinose, and vanillin. We present genetic evidence that the enzyme Rpd3 is the primary candidate for the deacetylation of Pab1-K131. Further, our electromobility shift assay studies suggest that the acetylation of Pab1-K131 negatively impacts poly(A) RNA binding. Due to the conserved nature of stress granules, therapeutics targeting the activity of lysine acetyltransferases and lysine deacetylase enzymes may be a promising route to modulate stress granule dynamics in the disease state.

Single-sequence protein structure prediction using supervised transformer protein language models

Significant progress has been made in protein structure prediction in recent years. However, it remains challenging for AlphaFold2 and other deep learning-based methods to predict protein structure with single-sequence input. Here we introduce trRosettaX-Single, an automated algorithm for single-sequence protein structure prediction. It incorporates the sequence embedding from a supervised transformer protein language model into a multi-scale network enhanced by knowledge distillation to predict inter-residue two-dimensional geometry, which is then used to reconstruct three-dimensional structures via energy minimization. Benchmark tests show that trRosettaX-Single outperforms AlphaFold2 and RoseTTAFold on orphan proteins and works well on human-designed proteins (with an average template modeling score (TM-score) of 0.79). An experimental test shows that the full trRosettaX-Single pipeline is two times faster than AlphaFold2, using much fewer computing resources (<10%). On 2,000 designed proteins from network hallucination, trRosettaX-Single generates structure models with high confidence. As a demonstration, trRosettaX-Single is applied to missense mutation analysis. These data suggest that trRosettaX-Single may find potential applications in protein design and related studies.

The High Mutational Sensitivity of ccdA Antitoxin Is Linked to Codon Optimality

Deep mutational scanning studies suggest that synonymous mutations are typically silent and that most exposed, nonactive-site residues are tolerant to mutations. Here, we show that the ccdA antitoxin component of the Escherichia coli ccdAB toxin-antitoxin system is unusually sensitive to mutations when studied in the operonic context. A large fraction (∼80%) of single-codon mutations, including many synonymous mutations in the ccdA gene shows inactive phenotype, but they retain native-like binding affinity towards cognate toxin, CcdB. Therefore, the observed phenotypic effects are largely not due to alterations in protein structure/stability, consistent with a large region of CcdA being intrinsically disordered. E. coli codon preference and strength of ribosome-binding associated with translation of downstream ccdB gene are found to be major contributors of the observed ccdA mutant phenotypes. In select cases, proteomics studies reveal altered ratios of CcdA:CcdB protein levels in vivo, suggesting that the ccdA mutations likely alter relative translation efficiencies of the two genes in the operon. We extend these results by studying single-site synonymous mutations that lead to loss of function phenotypes in the relBE operon upon introduction of rarer codons. Thus, in their operonic context, genes are likely to be more sensitive to both synonymous and nonsynonymous point mutations than inferred previously.

Intragenic compensation through the lens of deep mutational scanning

A significant fraction of mutations in proteins are deleterious and result in adverse consequences for protein function, stability, or interaction with other molecules. Intragenic compensation is a specific case of positive epistasis when a neutral missense mutation cancels effect of a deleterious mutation in the same protein. Permissive compensatory mutations facilitate protein evolution, since without them all sequences would be extremely conserved. Understanding compensatory mechanisms is an important scientific challenge at the intersection of protein biophysics and evolution. In human genetics, intragenic compensatory interactions are important since they may result in variable penetrance of pathogenic mutations or fixation of pathogenic human alleles in orthologous proteins from related species. The latter phenomenon complicates computational and clinical inference of an allele's pathogenicity. Deep mutational scanning is a relatively new technique that enables experimental studies of functional effects of thousands of mutations in proteins. We review the important aspects of the field and discuss existing limitations of current datasets. We reviewed ten published DMS datasets with quantified functional effects of single and double mutations and described rates and patterns of intragenic compensation in eight of them.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12551-022-01005-w.

Higher-order epistasis and phenotypic prediction

One core goal of genetics is to systematically understand the mapping between the DNA sequence of an organism (genotype) and its measurable characteristics (phenotype). Understanding this mapping is often challenging because of interactions between mutations, where the result of combining several different mutations can be very different than the sum of their individual effects. Here we provide a statistical framework for modeling complex genetic interactions of this type. The key idea is to ask how fast the effects of mutations change when introducing the same mutation in increasingly distant genetic backgrounds. We then propose a model for phenotypic prediction that takes into account this tendency for the effects of mutations to be more similar in nearby genetic backgrounds.

Contemporary high-throughput mutagenesis experiments are providing an increasingly detailed view of the complex patterns of genetic interaction that occur between multiple mutations within a single protein or regulatory element. By simultaneously measuring the effects of thousands of combinations of mutations, these experiments have revealed that the genotype–phenotype relationship typically reflects not only genetic interactions between pairs of sites but also higher-order interactions among larger numbers of sites. However, modeling and understanding these higher-order interactions remains challenging. Here we present a method for reconstructing sequence-to-function mappings from partially observed data that can accommodate all orders of genetic interaction. The main idea is to make predictions for unobserved genotypes that match the type and extent of epistasis found in the observed data. This information on the type and extent of epistasis can be extracted by considering how phenotypic correlations change as a function of mutational distance, which is equivalent to estimating the fraction of phenotypic variance due to each order of genetic interaction (additive, pairwise, three-way, etc.). Using these estimated variance components, we then define an empirical Bayes prior that in expectation matches the observed pattern of epistasis and reconstruct the genotype–phenotype mapping by conducting Gaussian process regression under this prior. To demonstrate the power of this approach, we present an application to the antibody-binding domain GB1 and also provide a detailed exploration of a dataset consisting of high-throughput measurements for the splicing efficiency of human pre-mRNA 5′ splice sites, for which we also validate our model predictions via additional low-throughput experiments.

Comprehensive fitness landscape of SARS-CoV-2 Mpro reveals insights into viral resistance mechanisms

With the continual evolution of new strains of severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) that are more virulent, transmissible, and able to evade current vaccines, there is an urgent need for effective anti-viral drugs. The SARS-CoV-2 main protease (Mpro) is a leading target for drug design due to its conserved and indispensable role in the viral life cycle. Drugs targeting Mpro appear promising but will elicit selection pressure for resistance. To understand resistance potential in Mpro, we performed a comprehensive mutational scan of the protease that analyzed the function of all possible single amino acid changes. We developed three separate high throughput assays of Mpro function in yeast, based on either the ability of Mpro variants to cleave at a defined cut-site or on the toxicity of their expression to yeast. We used deep sequencing to quantify the functional effects of each variant in each screen. The protein fitness landscapes from all three screens were strongly correlated, indicating that they captured the biophysical properties critical to Mpro function. The fitness landscapes revealed a non-active site location on the surface that is extremely sensitive to mutation, making it a favorable location to target with inhibitors. In addition, we found a network of critical amino acids that physically bridge the two active sites of the Mpro dimer. The clinical variants of Mpro were predominantly functional in our screens, indicating that Mpro is under strong selection pressure in the human population. Our results provide predictions of mutations that will be readily accessible to Mpro evolution and that are likely to contribute to drug resistance. This complete mutational guide of Mpro can be used in the design of inhibitors with reduced potential of evolving viral resistance.

Relation Between the Number of Peaks and the Number of Reciprocal Sign Epistatic Interactions

Empirical essays of fitness landscapes suggest that they may be rugged, that is having multiple fitness peaks. Such fitness landscapes, those that have multiple peaks, necessarily have special local structures, called reciprocal sign epistasis (Poelwijk et al. in J Theor Biol 272:141-144, 2011). Here, we investigate the quantitative relationship between the number of fitness peaks and the number of reciprocal sign epistatic interactions. Previously, it has been shown (Poelwijk et al. in J Theor Biol 272:141-144, 2011) that pairwise reciprocal sign epistasis is a necessary but not sufficient condition for the existence of multiple peaks. Applying discrete Morse theory, which to our knowledge has never been used in this context, we extend this result by giving the minimal number of reciprocal sign epistatic interactions required to create a given number of peaks.

Determinants of Multiheme Cytochrome Extracellular Electron Transfer Uncovered by Systematic Peptide Insertion

The multiheme cytochrome MtrA enables microbial respiration by transferring electrons across the outer membrane to extracellular electron acceptors. While structural studies have identified residues that mediate the binding of MtrA to hemes and to other cytochromes that facilitate extracellular electron transfer (EET), the relative importance of these interactions for EET is not known. To better understand EET, we evaluated how insertion of an octapeptide across all MtrA backbone locations affects Shewanella oneidensis MR-1 respiration on Fe(III). The EET efficiency was found to be inversely correlated with the proximity of the insertion to the heme prosthetic groups. Mutants with decreased EET efficiencies also arose from insertions in a subset of the regions that make residue-residue contacts with the porin MtrB, while all sites contacting the extracellular cytochrome MtrC presented high peptide insertion tolerance. MtrA variants having peptide insertions within the CXXCH motifs that coordinate heme cofactors retained some ability to support respiration on Fe(III), although these variants presented significantly decreased EET efficiencies. Furthermore, the fitness of cells expressing different MtrA variants under Fe(III) respiration conditions correlated with anode reduction. The peptide insertion profile, which represents the first comprehensive sequence-structure-function map for a multiheme cytochrome, implicates MtrA as a strategic protein engineering target for the regulation of EET.

From systems to structure - using genetic data to model protein structures

Understanding the effects of genetic variation is a fundamental problem in biology that requires methods to analyse both physical and functional consequences of sequence changes at systems-wide and mechanistic scales. To achieve a systems view, protein interaction networks map which proteins physically interact, while genetic interaction networks inform on the phenotypic consequences of perturbing these protein interactions. Until recently, understanding the molecular mechanisms that underlie these interactions often required biophysical methods to determine the structures of the proteins involved. The past decade has seen the emergence of new approaches based on coevolution, deep mutational scanning and genome-scale genetic or chemical-genetic interaction mapping that enable modelling of the structures of individual proteins or protein complexes. Here, we review the emerging use of large-scale genetic datasets and deep learning approaches to model protein structures and their interactions, and discuss the integration of structural data from different sources.

Epistatic drift causes gradual decay of predictability in protein evolution

Epistatic interactions can make the outcomes of evolution unpredictable, but no comprehensive data are available on the extent and temporal dynamics of changes in the effects of mutations as protein sequences evolve. Here, we use phylogenetic deep mutational scanning to measure the functional effect of every possible amino acid mutation in a series of ancestral and extant steroid receptor DNA binding domains. Across 700 million years of evolution, epistatic interactions caused the effects of most mutations to become decorrelated from their initial effects and their windows of evolutionary accessibility to open and close transiently. Most effects changed gradually and without bias at rates that were largely constant across time, indicating a neutral process caused by many weak epistatic interactions. Our findings show that protein sequences drift inexorably into contingency and unpredictability, but that the process is statistically predictable, given sufficient phylogenetic and experimental data.

Droplet-based screening of phosphate transfer catalysis reveals how epistasis shapes MAP kinase interactions with substrates

The combination of ultrahigh-throughput screening and sequencing informs on function and intragenic epistasis within combinatorial protein mutant libraries. Establishing a droplet-based, in vitro compartmentalised approach for robust expression and screening of protein kinase cascades (>10⁷ variants/day) allowed us to dissect the intrinsic molecular features of the MKK-ERK signalling pathway, without interference from endogenous cellular components. In a six-residue combinatorial library of the MKK1 docking domain, we identified 29,563 sequence permutations that allow MKK1 to efficiently phosphorylate and activate its downstream target kinase ERK2. A flexibly placed hydrophobic sequence motif emerges which is defined by higher order epistatic interactions between six residues, suggesting synergy that enables high connectivity in the sequence landscape. Through positive epistasis, MKK1 maintains function during mutagenesis, establishing the importance of co-dependent residues in mammalian protein kinase-substrate interactions, and creating a scenario for the evolution of diverse human signalling networks.

Here, the authors use a droplet-based screen for phosphate transfer catalysis, testing variants of the human protein kinase MKK1 for its ability to activate its downstream target ERK2. Data reveal a flexible motif in the MKK1 docking domain that promotes efficient activation of ERK2, and suggest epistasis between the residues within that sequence.

Modeling the ribosomal small subunit dynamic in Saccharomyces cerevisiae based on TCP-seq data

Translation Complex Profile Sequencing (TCP-seq), a protocol that was developed and implemented on Saccharomyces cerevisiae, provides the footprints of the small subunit (SSU) of the ribosome (with additional factors) across the entire transcriptome of the analyzed organism. In this study, based on the TCP-seq data, we developed for the first-time a predictive model of the SSU density and analyzed the effect of transcript features on the dynamics of the SSU scan in the 5′UTR. Among others, our model is based on novel tools for detecting complex statistical relations tailored to TCP-seq. We quantitatively estimated the effect of several important features, including the context of the upstream AUG, the upstream ORF length and the mRNA folding strength. Specifically, we suggest that around 50% of the variance related to the read counts (RC) distribution near a start codon can be attributed to the AUG context score. We provide the first large scale direct quantitative evidence that shows that indeed AUG context affects the small sub-unit movement. In addition, we suggest that strong folding may cause the detachment of the SSU from the mRNA. We also identified a number of novel sequence motifs that can affect the SSU scan; some of these motifs affect transcription factors and RNA binding proteins. The results presented in this study provide a better understanding of the biophysical aspects related to the SSU scan along the 5′UTR and of translation initiation in S. cerevisiae, a fundamental step toward a comprehensive modeling of initiation.

Learning protein fitness models from evolutionary and assay-labeled data

Machine learning-based models of protein fitness typically learn from either unlabeled, evolutionarily related sequences or variant sequences with experimentally measured labels. For regimes where only limited experimental data are available, recent work has suggested methods for combining both sources of information. Toward that goal, we propose a simple combination approach that is competitive with, and on average outperforms more sophisticated methods. Our approach uses ridge regression on site-specific amino acid features combined with one probability density feature from modeling the evolutionary data. Within this approach, we find that a variational autoencoder-based probability density model showed the best overall performance, although any evolutionary density model can be used. Moreover, our analysis highlights the importance of systematic evaluations and sufficient baselines.

Predicting and interpreting large-scale mutagenesis data using analyses of protein stability and conservation

Understanding and predicting the functional consequences of single amino acid changes is central in many areas of protein science. Here, we collect and analyze experimental measurements of effects of >150,000 variants in 29 proteins. We use biophysical calculations to predict changes in stability for each variant and assess them in light of sequence conservation. We find that the sequence analyses give more accurate prediction of variant effects than predictions of stability and that about half of the variants that show loss of function do so due to stability effects. We construct a machine learning model to predict variant effects from protein structure and sequence alignments and show how the two sources of information support one another and enable mechanistic interpretations. Together, our results show how one can leverage large-scale experimental assessments of variant effects to gain deeper and general insights into the mechanisms that cause loss of function.

Heterogeneity of the GFP fitness landscape and data-driven protein design

Studies of protein fitness landscapes reveal biophysical constraints guiding protein evolution and empower prediction of functional proteins. However, generalisation of these findings is limited due to scarceness of systematic data on fitness landscapes of proteins with a defined evolutionary relationship. We characterized the fitness peaks of four orthologous fluorescent proteins with a broad range of sequence divergence. While two of the four studied fitness peaks were sharp, the other two were considerably flatter, being almost entirely free of epistatic interactions. Mutationally robust proteins, characterized by a flat fitness peak, were not optimal templates for machine-learning-driven protein design - instead, predictions were more accurate for fragile proteins with epistatic landscapes. Our work paves insights for practical application of fitness landscape heterogeneity in protein engineering.

Neural networks to learn protein sequence-function relationships from deep mutational scanning data

The mapping from protein sequence to function is highly complex, making it challenging to predict how sequence changes will affect a protein's behavior and properties. We present a supervised deep learning framework to learn the sequence-function mapping from deep mutational scanning data and make predictions for new, uncharacterized sequence variants. We test multiple neural network architectures, including a graph convolutional network that incorporates protein structure, to explore how a network's internal representation affects its ability to learn the sequence-function mapping. Our supervised learning approach displays superior performance over physics-based and unsupervised prediction methods. We find that networks that capture nonlinear interactions and share parameters across sequence positions are important for learning the relationship between sequence and function. Further analysis of the trained models reveals the networks' ability to learn biologically meaningful information about protein structure and mechanism. Finally, we demonstrate the models' ability to navigate sequence space and design new proteins beyond the training set. We applied the protein G B1 domain (GB1) models to design a sequence that binds to immunoglobulin G with substantially higher affinity than wild-type GB1.

AMaLa: Analysis of Directed Evolution Experiments via Annealed Mutational Approximated Landscape

We present Annealed Mutational approximated Landscape (AMaLa), a new method to infer fitness landscapes from Directed Evolution experiments sequencing data. Such experiments typically start from a single wild-type sequence, which undergoes Darwinian in vitro evolution via multiple rounds of mutation and selection for a target phenotype. In the last years, Directed Evolution is emerging as a powerful instrument to probe fitness landscapes under controlled experimental conditions and as a relevant testing ground to develop accurate statistical models and inference algorithms (thanks to high-throughput screening and sequencing). Fitness landscape modeling either uses the enrichment of variants abundances as input, thus requiring the observation of the same variants at different rounds or assuming the last sequenced round as being sampled from an equilibrium distribution. AMaLa aims at effectively leveraging the information encoded in the whole time evolution. To do so, while assuming statistical sampling independence between sequenced rounds, the possible trajectories in sequence space are gauged with a time-dependent statistical weight consisting of two contributions: (i) an energy term accounting for the selection process and (ii) a generalized Jukes-Cantor model for the purely mutational step. This simple scheme enables accurately describing the Directed Evolution dynamics and inferring a fitness landscape that correctly reproduces the measures of the phenotype under selection (e.g., antibiotic drug resistance), notably outperforming widely used inference strategies. In addition, we assess the reliability of AMaLa by showing how the inferred statistical model could be used to predict relevant structural properties of the wild-type sequence.

ECNet is an evolutionary context-integrated deep learning framework for protein engineering

Machine learning has been increasingly used for protein engineering. However, because the general sequence contexts they capture are not specific to the protein being engineered, the accuracy of existing machine learning algorithms is rather limited. Here, we report ECNet (evolutionary context-integrated neural network), a deep-learning algorithm that exploits evolutionary contexts to predict functional fitness for protein engineering. This algorithm integrates local evolutionary context from homologous sequences that explicitly model residue-residue epistasis for the protein of interest with the global evolutionary context that encodes rich semantic and structural features from the enormous protein sequence universe. As such, it enables accurate mapping from sequence to function and provides generalization from low-order mutants to higher-order mutants. We show that ECNet predicts the sequence-function relationship more accurately as compared to existing machine learning algorithms by using ~50 deep mutational scanning and random mutagenesis datasets. Moreover, we used ECNet to guide the engineering of TEM-1 β-lactamase and identified variants with improved ampicillin resistance with high success rates.

Saturation Mutagenesis of the Transmembrane Region of HokC in Escherichia coli Reveals Its High Tolerance to Mutations

Cells adapt to different stress conditions, such as the antibiotics presence. This adaptation sometimes is achieved by changing relevant protein positions, of which the mutability is limited by structural constrains. Understanding the basis of these constrains represent an important challenge for both basic science and potential biotechnological applications. To study these constraints, we performed a systematic saturation mutagenesis of the transmembrane region of HokC, a toxin used by Escherichia coli to control its own population, and observed that 92% of single-point mutations are tolerated and that all the non-tolerated mutations have compensatory mutations that reverse their effect. We provide experimental evidence that HokC accumulates multiple compensatory mutations that are found as correlated mutations in the HokC family multiple sequence alignment. In agreement with these observations, transmembrane proteins show higher probability to present correlated mutations and are less densely packed locally than globular proteins; previous mutagenesis results on transmembrane proteins further support our observations on the high tolerability to mutations of transmembrane regions of proteins. Thus, our experimental results reveal the HokC transmembrane region high tolerance to loss-of-function mutations that is associated with low sequence conservation and high rate of correlated mutations in the HokC family sequences alignment, which are features shared with other transmembrane proteins.