Biology subjects

Aris-Brosou, S.

Publications and source records attributed to Aris-Brosou, S..

At least 19 recordsLinked to original sources

bioRxiv2026

A moving target: non-stationary selection governs unsupervised prediction of viral fitness

Anticipating how mutations change viral fitness is central to genomic surveillance and vaccine design, yet the supervised phenotype data behind the most accurate variant-effect predictors are unavailable for most emerging pathogens. We ask how far label-free scoring can go using only sequences, their evolutionary history, and structure. We assemble a modular, fully unsupervised pipeline that estimates a few interpretable terms (intrinsic replicative fitness, antigenic escape, and realized growth), and that lets each term be produced by more than one estimator, so the estimator itself becomes a testable modeling choice. Benchmarking the intrinsic term on 21 viral deep-mutational-scanning assays from ProteinGym, we find that a 650-million-parameter single-sequence protein language model predicts viral mutational fitness weakly and heterogeneously (mean Spearman 0.15), whereas a trivial site-independent alignment model more than doubles it (0.39, better on 17 of 21 assays), with the largest gains on the antigenic surface proteins where the language model fails. Yet the ordering reverses across 186 non-viral ProteinGym assays, where the language model instead exceeds the alignment model, localizing the weakness to viral families under-represented in the model's training data. Alignment-conditioned language models (MSA Transformer, Tranception) recover this accuracy but do not clearly exceed the simple alignment, so the decisive feature is the family alignment, not model scale or architecture. Our central result is evolutionary. Using dated samples of SARS-CoV-2 spike and influenza H3N2 hemagglutinin, we show that the epoch of the alignment is itself a leading, virus-specific determinant of accuracy. This traces to non-stationary selection: the site-specific amino-acid preferences drift over time, abruptly for spike at the emergence of Omicron and gradually for H3N2 hemagglutinin. A phylogenetic mutation-selection estimator does not match the far cheaper alignment model, falling significantly below it on matched data. Unsupervised viral fitness prediction is, then, as much an evolutionary problem as a modeling one.

bioinformatics↗

bioRxiv2026

Unsupervised detection of antimicrobial-resistance determinants by coupling protein-language-models and evolutionary signatures

Antimicrobial resistance (AMR) is among the most pressing threats to global health, yet our ability to find resistance determinants is largely confined to what reference databases already contain: homology search and supervised classifiers recognize variants of known genes but are, by construction, blind to the larger environmental and clinical reservoir of determinants that have not yet been catalogued. To address this critical limitation, we tested whether resistance determinants can be flagged \emph{without} using any resistance label, by leveraging evolutionary signatures that acquisition and adaptation leave in bacterial genomes. We describe a label-free, multi-view framework that scores every gene family of a pangenome on five orthogonal axes: protein-language-model novelty relative to known protein space, mobility/compositional anomaly, episodic positive selection, presence/ absence homoplasy, and reconciliation-inferred horizontal transfer. These views were then combined based on a conjunctive (weighted geometric-mean) rule, so that only families implicated by several independent lines of evolutionary evidence score high. On a controlled simulation the conjunction recovers all planted determinants where no single view is specific. Applied without retraining to the Escherichia coli (n=150) and Klebsiella pneumoniae (n=150) pangenomes, the results confirm that known determinants are almost entirely accessory and concentrates them near the top of the ranking for K. pneumoniae (6.6-fold enrichment in the top 1%), but not for E. coli. Ablation shows presence/absence homoplasy carries most of the signal, that episodic selection is counterproductive, and that an equal-weighted conjunction is suboptimal. The framework offers a reproducible, database-independent shortlist of candidate determinants and a honest accounting of where evolutionary signal is, but is not sufficient on its own.

bioinformatics↗

bioRxiv2026

Glacial meltwater drives gene-specific diversification of metal resistance genes in High Arctic soil microbiomes

Climate warming accelerates glacial meltwater delivery to Arctic lakes, mobilizing metals from thawing catchments and reshaping the selective landscape experienced by resident microbes. Whether these gradients leave detectable evolutionary signatures in environmental resistance genes remains unclear. We investigated four metal resistance genes (merA, arsC, cadA, and chrR) in metagenomic datasets from Lake Hazen (Nunavut, Canada), the largest High Arctic freshwater lake, sampled across a natural hydrological gradient of Control, Low-runoff, and High-runoff regimes. Using a space-for-time design, we combined population-genetic and codon-based approaches to quantify diversity and selection, including nucleotide diversity, Tajimas D, nonsynonymous/synonymous diversity ratios, McDonald-Kreitman tests, site-level episodic selection, and Bayesian coalescent and structured-coalescent inference of gene-pool diversity and exchange. This multi-pronged approach allowed us to distinguish gene-specific evolutionary responses from broader demographic effects. We found marked heterogeneity among genes: merA showed increasing diversity and consistent evidence of adaptive evolution along the runoff gradient; cadA displayed the strongest adaptive signal under low runoff, with selective constraint patterns that varied across regimes; chrR exhibited the clearest signature of episodic positive selection, with gene-wide selection and adaptive substitutions concentrated in the high-runoff regime where chromium concentrations were greatest; while arsC remained largely consistent with neutral evolution across regimes. Together, these results show that climate-driven metal mobilization is associated with gene- and regime-specific diversification of microbial metal resistance gene pools in Arctic soil microbiomes. Differences in metal speciation (with arsenic and chromium occurring as redox-sensitive oxyanions, and mercury, cadmium and zinc as divalent cations) may contribute to the contrasting evolutionary trajectories observed across genes, although in-situ speciation was not assessed in this dataset. Our findings highlight environmental resistance genes as sensitive indicators of changing biogeochemical conditions in rapidly warming polar ecosystems.

microbiology↗

bioRxiv2026

Variant emergence, not vaccine deployment, drives episodic positive selection on the SARS-CoV-2 spike at provincial scale in Canada

Mass immunization against SARS-CoV-2 created a heterogeneous landscape of antibody-mediated immune pressure, yet whether this pressure measurably altered episodic positive selection on spike remains unresolved. Using Canadian genomic surveillance data spanning the five major variants of concern (Alpha, Beta, Gamma, Delta, and Omicron), we inferred time-resolved phylogenies from spike-coding sequences and applied site- and branch-level episodic selection models to identify when and where adaptive change occurred. To evaluate whether vaccination intensity was associated with selection, we integrated these phylogenetic analyses with provincial vaccination time series using cross-correlation and lagged panel regression models that accounted for province and time effects, lineage prevalence, and sampling heterogeneity. Episodic positive selection was concentrated at a limited number of spike codons, especially within the N-terminal domain, receptor-binding domain, and furin cleavage region. However, these signals were dominated by substitutions associated with variant emergence, particularly during the Alpha-to-Delta transition, rather than by vaccination rollout. Whole-gene tests provided no evidence that vaccine intensity was associated with elevated episodic selection, and residualized vaccination trajectories did not predict selection at biologically plausible lags. Across provinces, the timing and distribution of selection events were inconsistent with a vaccine-driven escape model. Together, these results indicate that, at provincial resolution in Canada, episodic positive selection on SARS-CoV-2 spike was driven primarily by variant turnover rather than vaccine deployment. More broadly, this study provides a quantitative, VOC-resolved assessment of spike evolution in a structured epidemic and suggests that population-level vaccination intensity was not a detectable determinant of spike adaptation in the period examined.

evolutionary biology↗

bioRxiv2026

dna-parser: a Python library written in Rust for fast encoding of DNA and RNA sequences

BackgroundThe ever-growing amount of available biological data leads modern analysis to be performed on large datasets. Unfortunately, bioinformatics tools for preprocessing and analyzing data are not always designed to treat such large amounts of data efficiently. Notably, this is the case when encoding DNA and RNA sequences into numerical representations, also called descriptors, before passing them to machine learning models. Furthermore, current Python tools available for this preprocessing step are not well suited to be integrated into pipelines resulting in slow encoding speeds. ResultsWe introduce dna-parser, a Python library written in Rust to encode DNA and RNA sequences into numerical features. The combination of Rust and Python allows to encode sequences rapidly and in parallel across multiple threads while maintaining compatibility with packages from the Python ecosystem. Moreover, this library implements many of the most widely used types of numerical feature schemes coming from bioinformaticss and natural language processing. Conclusiondna-parser is an easy to install Python library that offers many Python wheels for Linux (muslinux and manylinux), macOS, and Windows via pip (https://pypi.org/project/dna-parser/). The open source code is available on GitHub (https://github.com/Mvila035/dna_parser) along with the documentation (https://mvila035.github.io/dna_parser/documentation/).

bioinformatics↗

bioRxiv2025

Polar ecosystems harbor highly divergent DNA viruses underrepresented in current reference databases

Viruses are the most abundant biological entities on Earth, yet most of their diversity remains unsampled and unevenly represented across ecosystems. To test whether viral novelty is structured more by biome or latitude, we analyzed 120 curated environmental DNA metagenomes spanning marine and terrestrial habitats across equatorial temperate, Arctic, and Antarctic regions. Using a unified metagenomic and phylogenomic workflow, we reconstructed viral protein families, linked environmental sequences to the IMGVR v4.1 reference collection, and quantified divergence using three complementary measures: within-clade patristic distance, the bridging edge to the nearest reference neighbor, and phylogenetic tree-shape statistics. Because distances within phylogenetic trees are not independent, all inference was performed at the sample level. Across all samples, reconstructed environmental viral sequences were substantially more divergent from current reference genomes than their database homologs. The dominant pattern was biome-driven: terrestrial viromes were consistently farther from IMGVR references than marine viromes in every latitudinal band, and this contrast was supported by both branch-length and topology-based metrics. Latitude modulated this baseline signal, with polar ecosystems, especially Antarctic terrestrial samples, occupying the most divergent end of the two-axis landscape. Tree imbalance statistics showed that terrestrial viral phylogenies were more deeply branching and more asymmetric than marine ones, providing independent support for the biome effect. In contrast, some previously reported polar contrasts weakened after correcting for pseudoreplication and using proper sample-level replication. These results argue that viral novelty is structured primarily by biome, with polar environments amplifying an already strong terrestrial-marine contrast. The findings also indicate that current reference databases remain disproportionately sparse for terrestrial, especially polar terrestrial, viromes. Expanding viral sampling in these environments will be essential for improving ecological inference, improving database coverage, and refining our understanding of global viral diversity.

evolutionary biology↗

bioRxiv2025

Global warming drives the evolutionary rate of H1N1 and H3N2 influenza viruses

The haemagglutinin (HA) and neuraminidase (NA) genes of seasonal influenza A evolve under continual immune-driven positive selection. To test whether the tempo of selection has changed over time, we mapped branch- and site-specific episodic diversifying selection (MEME) onto Bayesian relaxed-clock time trees for HA and NA in H1N1 and H3N2, across multiple countries and four sequence-subsampling schemes. We dated each selection episode and tested whether episodes accumulated through time after accounting for the growing number of sampled lineages. Positive-selection episodes increased over time in every gene-subtype combination, at about 2-6% per lineage-year, and rose faster for NA than HA. Episodes were concentrated at a small number of codon sites, especially recurrent sites in H3N2 HA that fell within canonical antigenic regions of the HA1 head. This increase was robust to subsampling scheme and time-bin width, and was driven disproportionately by recent lineages. A detrended spatial analysis found no association with latitude or temperature anomalies. Overall, positive selection on influenza surface antigens appears to be intensifying through time, most likely because of immune escape and expanded surveillance rather than climate warming.

evolutionary biology↗

bioRxiv2024

The High Arctic is dominated by uncharacterized, genetically highly diverse bacteriophages

While the Earths virosphere is estimated to be in the range of 1031 viral particles, the vast majority of its diversity has yet to be discovered. In recent years, metagenomics has rapidly allowed the identification of viruses, from microenvironments to extreme environments like the High Arctic. However, the High Arctic virome is largely composed of viral sequences that have few matches to classified viruses in existing databases. Here, to bypass limitations posed by similarity-based strategies, we resorted to a paired metagenomic (DNA, n = 8) and metatranscriptomic (RNA, n = 6) approach that placed viral genes found in Lake Hazen, the worlds largest High Arctic lake, in a phylogenetic context with known viruses. We show that High Arctic viruses have undergone unique evolutionary processes characterized by high evolutionary rates, making them distinct from and more diverse than known viruses.

evolutionary biology↗

bioRxiv2021

Viral spillover risk in High Arctic increases with melting glaciers

The host spectrum of viruses is quite diverse, as they can sustainedly infect a few species to several phyla. When confronted with a new host, a virus may even infect it and transmit sustainably in this new host, a process called "viral spillover." However, the risk of such events is difficult to quantify. As climate change is rapidly transforming environments, it is becoming critical to quantify the potential for spillovers. To address this issue, we resorted to a metagenomics approach and focused on two environments, soil and lake sediments from Lake Hazen, the largest High Arctic freshwater lake in the world. We used DNA and RNA sequencing to reconstruct the lakes virosphere in both its sediments and soils, as well as its range of eukaryotic hosts. We then estimated the spillover risk by measuring the congruence between the viral and the eukaryotic host phylogenetic trees, and show that spillover risk is higher in lake sediments than in soil, and increases with runoff from glacier melt, a proxy for climate change. Should climate change also shift species range of potential viral vectors and reservoirs northwards, the High Arctic could become fertile ground for emerging pandemics.

ecology↗

bioRxiv2021

Viral evolutionary dynamics predict Influenza-Like-Illnesses in patients

Viral infections such as those caused by the influenza virus can put a strain on healthcare systems. However, such a burden is typically difficult to predict. In order to improve such predictions, we hypothesize that the severity of epidemics can be linked to viral evolutionary dynamics. More specifically, we posit the existence of a negative association between patients health and the stability of coevolutionary networks at key viral proteins. To test this, we performed a thorough evolutionary analysis of influenza viruses circulating in continental US between 2010 and 2019, assessing how measures of the stability of these coevolutionary networks correlate with clinical data based on outpatient healthcare visits showing Influenza-Like Illness (ILI) symptoms. We first show evidence of a significant correlation between viral evolutionary dynamics and increased influenza activity during seasonal epidemics, and then show that these dynamics closely follow the progression of epidemics through each season, providing us with predictive power based on genetic data collected between week 20 and week 40/52, that is one to fifteen weeks prior to peak ILI. Viral evolutionary dynamics may hence be used by health authorities to further guide non-pharmaceutical interventions.

evolutionary biology↗

bioRxiv2019

Microbial genomes retrieved from High Arctic lake sediments provision their microbes with genes to strive in cold and oligotrophic environments

The Arctic is currently warming at an unprecedented rate, which may affect environmental constraints on the freshwater microbial communities found there. Yet, our knowledge of the community structure and functional potential of High Arctic freshwater microbes remains poor, even though they play key roles in nutrient cycling and other ecosystem services. Here, using high-throughput metagenomic sequencing and genome assembly, we show that sediment microbial communities in the High Arctics largest lake by volume, Lake Hazen, are phylogenetically diverse, ranging from Proteobacteria, Verrucomicrobia, Planctomycetes, to members of the newly discovered Candidate Phyla Radiation (CPR) groups. These genomes displayed a high prevalence of pathways involved in lipid chemistry, and a low prevalence of nutrient uptake pathways, which might represent adaptations to the specific, cold (~3.5{degrees}C) and extremely oligotrophic conditions in Lake Hazen. Despite these potential adaptations, it is unclear how ongoing environmental changes will affect microbial communities, the makeup of their genomic idiosyncrasies, as well as the possible implications at higher trophic levels.

microbiology↗

bioRxiv2019

Climate change lowers diversity and functional potential of microbes in Canada’s high Arctic

Temperatures in the Arctic are expected to increase dramatically over the next century, yet little is known about how microbial communities and their underlying metabolic processes will be affected by these environmental changes in freshwater sedimentary systems. To address this knowledge gap, we analyzed sediments from Lake Hazen, NU Canada. Here, we exploit the spatial heterogeneity created by varying runoff regimes across the watershed of this uniquely large lake at these latitudes to test how a transition from low to high runoff, used as one proxy for climate change, affects the community structure and functional potential of dominant microbes. Based on metagenomic analyses of lake sediments along these spatial gradients, we show that increasing runoff leads to a decrease in taxonomic and functional diversity of sediment microbes. Our findings are likely to apply to other, smaller, glacierized watersheds typical of polar or high latitude / high altitudes ecosystems; we can predict that such changes will have far reaching consequences on these ecosystems by affecting nutrient biogeochemical cycling, the direction and magnitude of which are yet to be determined.

microbiology↗

bioRxiv2019

Episodic diversifying selection and intragenomic interactions shape the evolution of DNA and RNA viruses

Viruses are known to have some of the highest and most diverse mutation rates found in any biological replicator, topped by single-stranded (ss) RNA viruses, while double-stranded (ds) DNA viruses have rates approaching those of bacteria. As mutation rates are tightly and negatively correlated with genome size, selection is a clear driver of viral evolution. However, the role of intragenomic interactions as drivers of viral evolution is less well documented. To understand how these two processes affect viral evolution, we systematically surveyed ssRNA, ssDNA, dsRNA, and dsDNA viruses, to find which virus type and which functions show evidence for episodic diversifying selection and correlated evolution. We show that while evidence for selection is mostly found in single stranded viruses, and correlated evolution is more prevalent in DNA viruses, the genes that are affected by both processes are involved in key aspects of their life cycle, favoring viral stability over proliferation. We further show that both evolutionary processes are intimately linked at the amino acid level, which suggests that selection alone does not explain the whole evolutionary --and epidemiological-- potential of viruses.

evolutionary biology↗

bioRxiv2018

The idiosyncratic drivers of correlated evolution

The ultimate causes of correlated evolution among sites in a genome remain difficult to tease apart. To address this problem directly, we performed a high-throughput search for correlated evolution among sites associated with resistance to a fluoroquinolone antibiotic using whole genome data from clinical strains of Pseudomonas aeruginosa, before validating our computational predictions experimentally. We show that for at least two sites, this correlation is underlain by epistasis. Our analysis also revealed eight additional pairs of synonymous substitutions displaying correlated evolution underlain by physical linkage, rather than selection associated with antibiotic resistance. Our results provide direct evidence that both epistasis and physical linkage among sites can drive the correlated evolution identified by high throughput computational tools. In other words, the observation of correlated evolution is not by itself sufficient evidence to guarantee that the sites in question are epistatic; such a claim requires additional evidence, ideally coming from direct estimates of epistasis, based on experimental evidence.

bioinformatics↗

bioRxiv2018

Evidence of a nonadaptive buildup of mutational load in human populations over the past 40,000 years

The extent to which selection has shaped present-day human populations has attracted intense scrutiny, and examples of local adaptations abound. However, the evolutionary trajectory of alleles that, today, are deleterious has received much less attention. To address this question, the genomes of 2,062 individuals, including 1,179 ancient humans, were reanalyzed to assess how frequencies of risk alleles and their homozygosity changed through space and time in Europe over the past 45,000 years. While the overall deleterious homozygosity has consistently decreased, risk alleles have steadily increased in frequency over that period of time. Those that increased most are associated with diseases such as asthma, Crohn disease, diabetes and obesity, which are highly prevalent in present-day populations. These findings may not run against the existence of local adaptations, but highlight the limitations imposed by drift and population dynamics on the strength of selection in purging deleterious mutations from human populations.

evolutionary biology↗

bioRxiv2018

How the Central American Seaway and an ancient northern passage affected Flatfish diversification

While the natural history of flatfish has been debated for decades, the mode of diversification of this biologically and economically important group has never been elucidated. To address this question, we assembled the largest molecular data set to date, covering > 300 species (out of ca. 800 extant), from 13 of the 14 known families over nine genes, and employed relaxed molecular clocks to uncover their patterns of diversification. As the fossil record of flatfish is contentious, we used sister species distributed on both sides of the American continent to calibrate clock models based on the closure of the Central American Seaway (CAS), and on their current species range. We show that flatfish diversified in two bouts, as species that are today distributed around the Equator diverged during the closure of CAS, while those with a northern range diverged after this, hereby suggesting the existence of a post-CAS closure dispersal for these northern species, most likely along a trans-Arctic northern route, a hypothesis fully compatible with paleogeographic reconstructions.

evolutionary biology↗

bioRxiv2017

Identifying the genetic determinants of particular phenotypes in microbial genomes with very small training sets

Machine learning (ML) encompasses numerous algorithms that aim at discovering complex patterns between elements within large data using limited prior assumptions or modeling. However, some scientific disciplines still produce small data sets: in particular, empirical studies that try to find the mutations responsible for complex phenotypes are often limited to very small sample sizes (n), while scanning a large number of amino acid sites (p) in a proteome. To date, little is known on how ML performs in this type of so-called \"large p, small n\" problem. To address this question, we evaluated the performance of two general ML classifiers, adaptive boosting (AB) and random forest, on two data sets. To assess the impact of proteome size, we contrasted a small (viral) genome with a larger (bacterial) one. To analyze large proteomes, we further developed a chunking algorithm, and introduce a repeated random forest (RRF) algorithm that stabilizes model predictions. With the influenza data, we were able to rediscover amino acid sites experimentally implicated in three different complex phenotypes (infectivity, transmissibility, and pathogenicity). Results for the larger proteome, pertaining to three types of drug resistance (Ciprofloxacin, Ceftazidime, and Gentamicin), were more nuanced, with RRF making more sensible pre-dictions, with smaller errors rates, than AB. Furthermore, we show that chunking improved runtimes by an order of magnitude and may increase sensitivity of the predictions. Altogether, we demonstrate that ML algorithms can be used to identify genetic determinants in small proteomes (viruses), even with small numbers of individuals. We further show that even if the size of bacterial proteomes pushes AB to its limits in the context of small n, RRF may deserve more scrutiny, which should be facilitated by the plummeting costs of sequencing and, more critically, by phenotyping large cohorts of individuals.\n\nAuthor SummaryFinding the genetic determinants of a phenotype is typically performed by testing for an association between a particular allele and a trait, carrying out the testing over a large number of loci in a large cohort of individuals, itself divided into two subsets of individuals: those who have the trait (cases), and those who do not (controls). However, recruiting large cohorts can be problematic in some experimental fields, while using genotypic information rather than complete genomes can miss some mutations. To address these issues, we implemented two machine learning (ML) algorithms, tweaked for analyzing large genomes and providing stable results. The analysis of a small viral genome, for which genetic determinants of three phenotypes are already known, showed that our approach can rediscover known mutations, almost irrespective of the ML algorithm used. However, the analysis of a larger bacterial genome, for which genetic determinants of three phenotypes are unknown, suggested that the simpler of our modified algorithms performed better, returning more sensitive predictions with lower error rates. This work demonstrates the feasibility of finding genetic determinants of complex phenotypes based on a small number of complete genomes.

bioinformatics↗

bioRxiv2017

Approximate Bayesian Computation Algorithms for Estimating Network Model Parameters

Studies on Approximate Bayesian Computation (ABC) replacing the intractable likelihood function in evaluation of the posterior distribution have been developed for several years. However, their field of application has to date essentially been limited to inference in population genetics. Here, we propose to extend this approach to estimating the structure of transmission networks of viruses in human populations. In particular, we are interested in estimating the transmission parameters under four very general network structures: random, Watts-Strogatz, Barabasi-Albert and an extension that incorporates aging. Estimation was evaluated under three approaches, based on ABC, ABC-Markov chain Monte Carlo (ABC-MCMC) and ABC-Sequential Monte Carlo (ABC-SMC) samplers. We show that ABC-SMC samplers outperform both ABC and ABC-MCMC, achieving high accuracy and low variance in simulations. This approach paves the way to estimating parameters of real transmission networks of transmissible diseases.

evolutionary biology↗