SARS-CoV-2 variants evolve convergent strategies to remodel the host response.

SARS-CoV-2 variants of concern (VOCs) emerged during the COVID-19 pandemic. Here, we used unbiased systems approaches to study the host-selective forces driving VOC evolution. We discovered that VOCs evolved convergent strategies to remodel the host by modulating viral RNA and protein levels, altering viral and host protein phosphorylation, and rewiring virus-host protein-protein interactions. Integrative computational analyses revealed that although Alpha, Beta, Gamma, and Delta ultimately converged to suppress interferon-stimulated genes (ISGs), Omicron BA.1 did not. ISG suppression correlated with the expression of viral innate immune antagonist proteins, including Orf6, N, and Orf9b, which we mapped to specific mutations. Later Omicron subvariants BA.4 and BA.5 more potently suppressed innate immunity than early subvariant BA.1, which correlated with Orf6 levels, although muted in BA.4 by a mutation that disrupts the Orf6-nuclear pore interaction. Our findings suggest that SARS-CoV-2 convergent evolution overcame human adaptive and innate immune barriers, laying the groundwork to tackle future pandemics.

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.

Towards a structurally resolved human protein interaction network.

Cellular functions are governed by molecular machines that assemble through protein-protein interactions. Their atomic details are critical to studying their molecular mechanisms. However, fewer than 5% of hundreds of thousands of human protein interactions have been structurally characterized. Here we test the potential and limitations of recent progress in deep-learning methods using AlphaFold2 to predict structures for 65,484 human protein interactions. We show that experiments can orthogonally confirm higher-confidence models. We identify 3,137 high-confidence models, of which 1,371 have no homology to a known structure. We identify interface residues harboring disease mutations, suggesting potential mechanisms for pathogenic variants. Groups of interface phosphorylation sites show patterns of co-regulation across conditions, suggestive of coordinated tuning of multiple protein interactions as signaling responses. Finally, we provide examples of how the predicted binary complexes can be used to build larger assemblies helping to expand our understanding of human cell biology.

A structural biology community assessment of AlphaFold2 applications.

Most proteins fold into 3D structures that determine how they function and orchestrate the biological processes of the cell. Recent developments in computational methods for protein structure predictions have reached the accuracy of experimentally determined models. Although this has been independently verified, the implementation of these methods across structural-biology applications remains to be tested. Here, we evaluate the use of AlphaFold2 (AF2) predictions in the study of characteristic structural elements; the impact of missense variants; function and ligand binding site predictions; modeling of interactions; and modeling of experimental structural data. For 11 proteomes, an average of 25% additional residues can be confidently modeled when compared with homology modeling, identifying structural features rarely seen in the Protein Data Bank. AF2-based predictions of protein disorder and complexes surpass dedicated tools, and AF2 models can be used across diverse applications equally well compared with experimentally determined structures, when the confidence metrics are critically considered. In summary, we find that these advances are likely to have a transformative impact in structural biology and broader life-science research.

Exploring amino acid functions in a deep mutational landscape.

Amino acids fulfil a diverse range of roles in proteins, each utilising its chemical properties in different ways in different contexts to create required functions. For example, cysteines form disulphide or hydrogen bonds in different circumstances and charged amino acids do not always make use of their charge. The repertoire of amino acid functions and the frequency at which they occur in proteins remains understudied. Measuring large numbers of mutational consequences, which can elucidate the role an amino acid plays, was prohibitively time-consuming until recent developments in deep mutational scanning. In this study, we gathered data from 28 deep mutational scanning studies, covering 6,291 positions in 30 proteins, and used the consequences of mutation at each position to define a mutational landscape. We demonstrated rich relationships between this landscape and biophysical or evolutionary properties. Finally, we identified 100 functional amino acid subtypes with a data-driven clustering analysis and studied their features, including their frequencies and chemical properties such as tolerating polarity, hydrophobicity or being intolerant of charge or specific amino acids. The mutational landscape and amino acid subtypes provide a foundational catalogue of amino acid functional diversity, which will be refined as the number of studied protein positions increases.

Population Structure, Stratification, and Introgression of Human Structural Variation.

Structural variants contribute substantially to genetic diversity and are important evolutionarily and medically, but they are still understudied. Here we present a comprehensive analysis of structural variation in the Human Genome Diversity panel, a high-coverage dataset of 911 samples from 54 diverse worldwide populations. We identify, in total, 126,018 variants, 78% of which were not identified in previous global sequencing projects. Some reach high frequency and are private to continental groups or even individual populations, including regionally restricted runaway duplications and putatively introgressed variants from archaic hominins. By de novo assembly of 25 genomes using linked-read sequencing, we discover 1,643 breakpoint-resolved unique insertions, in aggregate accounting for 1.9 Mb of sequence absent from the GRCh38 reference. Our results illustrate the limitation of a single human reference and the need for high-quality genomes from diverse populations to fully discover and understand human genetic variation.

The impact of the genetic background on gene deletion phenotypes in Saccharomyces cerevisiae.

Loss-of-function (LoF) mutations associated with disease do not manifest equally in different individuals. The impact of the genetic background on the consequences of LoF mutations remains poorly characterized. Here, we systematically assessed the changes in gene deletion phenotypes for 3,786 gene knockouts in four Saccharomyces cerevisiae strains and 38 conditions. We observed 18.5% of deletion phenotypes changing between pairs of strains on average with a small fraction conserved in all four strains. Conditions causing higher wild-type growth differences and the deletion of pleiotropic genes showed above-average changes in phenotypes. In addition, we performed a genome-wide association study (GWAS) for growth under the same conditions for a panel of 925 yeast isolates. Gene-condition associations derived from GWAS were not enriched for genes with deletion phenotypes under the same conditions. However, cases where the results were congruent indicate the most likely mechanism underlying the GWAS signal. Overall, these results show a high degree of genetic background dependencies for LoF phenotypes.

Inferring Fitness Effects from Time-Resolved Sequence Data with a Delay-Deterministic Model.

A common challenge arising from the observation of an evolutionary system over time is to infer the magnitude of selection acting upon a specific genetic variant, or variants, within the population. The inference of selection may be confounded by the effects of genetic drift in a system, leading to the development of inference procedures to account for these effects. However, recent work has suggested that deterministic models of evolution may be effective in capturing the effects of selection even under complex models of demography, suggesting the more general application of deterministic approaches to inference. Responding to this literature, we here note a case in which a deterministic model of evolution may give highly misleading inferences, resulting from the nondeterministic properties of mutation in a finite population. We propose an alternative approach that acts to correct for this error, and which we denote the delay-deterministic model. Applying our model to a simple evolutionary system, we demonstrate its performance in quantifying the extent of selection acting within that system. We further consider the application of our model to sequence data from an evolutionary experiment. We outline scenarios in which our model may produce improved results for the inference of selection, noting that such situations can be easily identified via the use of a regular deterministic model.