Biology subjects

Rokas, A.

Publications and source records attributed to Rokas, A..

At least 19 recordsLinked to original sources

bioRxiv2026

Functional diversification across yeast lineages of the nucleus-vacuole junction forming protein Nvj1

Nucleus-vacuole junctions (NVJs) in Saccharomyces cerevisiae serve as an inter-organellar hub for multiple cell processes, including lipid transport and biosynthesis, an intra-nuclear quality control mechanism, and piecemeal microautophagy of the nucleus. NVJs are formed by complexes between Vac8 in the vacuole membrane and Nvj1 in the nuclear envelope. Nvj1 also links the inner and outer nuclear membranes across the perinuclear lumen. Orthologs of Nvj1 had previously only been found in yeasts of the order Saccharomycetales, raising the possibility that NVJs are restricted to this clade. Using homology and synteny, we discovered scores of novel Nvj1 orthologs across 200 million years of evolution within the subphylum Saccharomycotina. Not all orthologs mediate NVJ formation when expressed in S. cerevisiae, and some lack a sequence motif necessary for this function. Furthermore, the sequence motif required for binding the oxysterol binding protein Osh1 and an associated novel sequence motif, both found only in the order Saccharomycetales, were independently lost in three different lineages. This collection suggests specific opportunities to explore evolutionary and functional adaptations of Nvj1s in ecologically and physiologically diverse yeasts.

cell biology↗

bioRxiv2026

Evolutionary origins of protein novelty across an entire yeast subphylum

Novel protein-coding sequences fuel molecular and cellular evolutionary innovations and frequently contribute to species-defining characteristics. They can originate either de novo from previously noncoding sequences or through extreme divergence of already coding ones. How frequently each mechanism occurs and how they shape the structural and functional potential of the resulting proteins remains unclear. Here, we conducted a broad computational investigation of genetic and protein novelty throughout the entire subphylum of Saccharomycotina yeasts. We detected more than 5,000 robust de novo genes across 332 species and compared them to more than 6,000 novel genes resulting from extreme sequence divergence, revealing two quantitatively similar but qualitatively distinct modes of evolution of novelty. A remarkable 40% of de novo proteins are predicted to localize to mitochondria compared to only 20% of divergent, with the latter also being substantially longer and more disordered. A detailed analysis of conservatively predicted tertiary structures of novel proteins shows that ''invention'' of new folds occurs more frequently through de novo emergence. We also illustrate cases of evolutionary ''re-invention'' of existing protein folds from noncoding sequences. Our work deepens our understanding of the origins and importance of novel proteins, opening new directions for further structural and functional characterization.

genomics↗

bioRxiv2026

Opportunistic pathogenicity in fungi can transcend species boundaries

Multiple fungal lineages have independently evolved the ability to opportunistically infect humans, imposing a major burden on public health. Opportunistic pathogenicity requires the confluence of pre-existing fungal traits that facilitate host colonization (e.g., the ability to grow at 37{degrees}C) and host immune filters that permit the survival of some colonizers (e.g., inborn errors of immunity). Numerous studies have shown that fungal pathogens can exhibit extensive strain-to-strain variation in the ability to cause disease. Moreover, it is also well established that non-pathogenic fungi can occasionally cause severe infections. Together, these observations raise the question: What differentiates opportunistic fungal pathogens from non-pathogens? To address this empirically, we compared phenotypic, metabolomic, and genomic variation between Aspergillus fumigatus, an organism responsible for more than 300,000 infections per year, and Aspergillus fischeri, a close relative that is not considered clinically relevant. By examining 26 phenotypic traits across 16 representative strains of A. fumigatus and 16 of A. fischeri, we found that infection-relevant traits measured under in vitro monoculture conditions showed species-specific distributions, whereas traits measured under in vitro coculture with murine macrophages overlapped. Strikingly, strains of the two species also overlapped in their virulence profiles in an immunocompromised murine model of pulmonary aspergillosis; three strains of A. fischeri exhibited lethality rates of >50% while two A. fumigatus strains were among the least virulent of all 32 strains tested. Consistent with this overlap, we could not associate virulence variation with the presence of specific genomic elements, phenotypic traits, or secondary metabolites. Our results raise the hypothesis that opportunistic pathogenicity can extend beyond the boundaries of individual species. We propose a conceptual model in which the opportunistic pathogenic potential of any fungal strain is the product of complex interactions among numerous genomic, ecological, and host immunity factors.

evolutionary biology↗

bioRxiv2026

Processed pseudogenes as dynamic substrates of vertebrate genome evolution

Pseudogenes, gene copies presumed nonfunctional, are widespread products of genome evolution, yet their retention under selection and functional significance across vertebrate diversity remain poorly understood. Here, by analyzing 244 high-quality, chromosome-scale genomes from the Vertebrate Genomes Project spanning seven major vertebrate lineages, we show that the most abundant class of pseudogenes, processed pseudogenes (retrocopies), is a dynamic substrate for evolutionary innovation rather than an inert relic. Retrocopy abundance varies by more than an order of magnitude across lineages, closely tracks autonomous retrotransposon content, and a considerable fraction retains intact open reading frames under purifying selection. We establish a two-stage model in which a conserved formation bias toward highly expressed housekeeping genes is followed by lineage-specific selective filtering that shapes distinct functional repertoires. Testing this model, we show that the mammalian X chromosome exports retrocopies to autosomes at significantly elevated rates enriched for functionally constrained copies, establishing meiotic sex chromosome inactivation as the selective driver. Furthermore, tumor suppressor gene retrocopies accumulate preferentially over oncogene retrocopies in large-bodied and long-lived mammalian lineages, identifying retrocopy-mediated tumor suppressor dosage expansion as a previously unrecognized genomic correlate of Petos paradox. Beyond cancer-related dynamics, retrocopy abundance itself correlates with key mammalian life-history traits, including brain mass, generation length, and reproductive timing, which suggests that retrocopy turnover is broadly coupled to organismal pace-of-life. These findings recast retrocopies as a major axis of vertebrate genome evolution and provide a comprehensive resource for studying gene duplicate innovation.

genomics↗

bioRxiv2025

The Macroevolution of Filamentation Morphology Across the Saccharomycotina Yeast Subphylum

Saccharomycotina yeasts are a highly diverse and widely distributed subphylum of ascomycete fungi that exhibit diversity in their asexual growth morphologies; the human commensal yeast Candida albicans exhibits both unicellular and filamentous growth, whereas the bakers yeast Saccharomyces cerevisiae grows unicellularly by budding. Filamentous growth (and the production of hyphae) can comprise linear and branched budding cells that do not undergo cell separation, termed pseudohyphae, or of tubular filaments with septa that perforate allowing movement of organelles, termed true hyphae. We integrated phenotypic, genomic, metabolic, and isolation environment data from 1,154 strains from 1,051 Saccharomycotina species to examine the variation and evolutionary history of filamentation across Saccharomycotina and determine the extent to which genes, metabolic traits, and environments can predict filamentation types. We found that 63.37% of strains can form filaments; 6.56% can form true hyphae, 42.40% can form pseudohyphae, and 14.39% can form both true hyphae and pseudohyphae. We found that the distributions of species that can produce true hyphae or that can filament were more strongly correlated with the yeast phylogeny than the distribution of species with pseudohyphae. Ancestral state reconstruction suggested that the ability to form true hyphae and/or pseudohyphae evolved several times in yeast evolution. We infer that most yeast ancestors likely produced pseudohyphae or lacked filaments, while the Saccharomycotina last common ancestor likely produced pseudohyphae but not true hyphae. Machine learning models trained on genomic and metabolic features predicted filament morphologies with about 70% accuracy. Connecting the evolution of Saccharomycotina morphologies to their genomic, metabolic, and ecological characteristics will enrich our understanding of the evolution of the diversity of lifestyles observed in this lineage.

evolutionary biology↗

bioRxiv2025

A trait syndrome ties cell morphology to glycolysis across the yeast subphylum

Traits that co-vary across species can provide fundamental insights into the trade-offs and constraints that govern their evolution. It was recently reported that glucose uptake rates (GUR) are inversely correlated with the cell surface area-to-volume (SA:V) ratio across 11 yeast species. Here we substantially expand this analysis to 282 species to test whether the GUR-SA:V correlation generalizes across the ancient Saccharomycotina yeast subphylum and to determine the contribution of shared evolutionary history to the co-variation of these two traits. Using regression models that account for co-variation due to phylogeny, we found that extracellular acidification rates (ECAR, which we used as a scalable proxy for GUR) had a weak, but significant, correlation with SA:V across Saccharomycotina. We found additional weak, but significant, correlations between ECAR with genome sizes and growth rates. Our findings largely agree with the recently reported correlations between GUR and SA:V ratio, but they also show that there are likely several other associated variables, including genome size. Specifically, yeasts that consume glucose faster tend to have lower SA:V, faster growth rates, and larger genomes. Our results suggest that a trait syndrome governs several metabolic, genomic, and morphological traits across the yeast subphylum.

evolutionary biology↗

bioRxiv2025

Population-Specific Transcriptomic Shifts Underlie Secondary Metabolic Diversification in Aspergillus flavus and the Domestication of Aspergillus oryzae

The genomic signatures of microbial domestication remain poorly understood within the context of natural population variation. Here, we demonstrate that Aspergillus oryzae, the filamentous fungus used in soy sauce production, shares more recent ancestry with a predominantly northern, largely non-aflatoxigenic population of Aspergillus flavus (population C). Strikingly, A. oryzae isolates also overlap with the recently described, clinically enriched A. flavus population D, suggesting the possibility of multiple domestication events. Although A. oryzae exhibits reduced virulence compared to A. flavus, all isolates tested retained pathogenicity in a zebrafish infection model. At the transcriptomic level, Aspergillus populations are significantly differentiated, with distinct responses to population density, indicating that population-specific transcriptomes adapt differently to ecological conditions. These differences extend beyond gene content and are not always explained by phylogenetic relationships, suggesting that phenotypic diversification occurs through the rapid reorganization of transcriptomic architectures. For example, A. oryzae displays significantly elevated expression of a module enriched for carbohydrate metabolism. Population-specific variation is also evident among secondary metabolite (SM) gene clusters. While A. oryzae shows markedly reduced expression of specific SM genes, particularly those involved in aflatoxin biosynthesis, this trend does not extend across the entire secondary metabolome. Using machine- learning-based gene regulatory network inference, we identified population-specific transcriptomic differences linked to distinct transcription factors, with evidence for both cis- and trans-acting regulatory divergence, but no changes in global regulators such as laeA. Together, these findings provide new insights into the domestication of A. oryzae, its global significance, and the microevolution of fungal secondary metabolic pathways. Significance StatementAspergillus flavus poses a significant threat to global agriculture by infecting crops and producing aflatoxin, a potent carcinogen that undermines food safety and international trade. The species is also a major human pathogen causing invasive fungal infections. Here, we demonstrate that A. flavus populations exhibit not only variations in their capacity to produce toxins but also in the way they regulate gene expression in response to environmental cues. These regulatory differences played a key role in the domestication of Aspergillus oryzae, a non-aflatoxigenic relative that has been safely used in food fermentation for centuries. Notably, while A. oryzae exhibits reduced aflatoxin biosynthetic gene expression, it expresses other secondary metabolite genes and retains limited opportunistic pathogenic potential. By revealing how transcriptomic rewiring drives population-level divergence, this work provides new insights into the evolution of toxin production, with broad implications for microbial domestication, crop protection, and global food security.

microbiology↗

bioRxiv2025

Convergent evolution of aerobic fermentation through divergent mechanisms acting on key shared glycolytic genes

As the tree of life becomes increasingly accessible to molecular investigations, describing mechanisms underlying evolutionary convergence and constraint will be crucial to understanding diversification. The lineage including the model yeast Saccharomyces cerevisiae evolved aerobic fermentation in part through an ancient whole genome duplication and retention of glycolytic genes. To evaluate the glycolytic rates across diverse yeasts, we developed and deployed an extracellular acidification rates (ECAR) assay on 299 species that span more than 400 million years of evolution and identified a clade in the genus Saturnispora that convergently evolved aerobic fermentation. Through comparative genomics and transcriptomics, we found that several glycolytic genes had higher expression and novel cis-regulatory elements in aerobically fermenting Saturnispora species. When the transcription factor required for their activation was deleted in Saturnispora dispora, the mutants had reduced glycolytic rates and increased respiration. Intriguingly, many of the upregulated genes are orthologous to duplicated glycolytic genes in S. cerevisiae. These divergent genetic mechanisms suggest that there are strong evolutionary constraints on how some traits like aerobic fermentation can arise convergently.

evolutionary biology↗

bioRxiv2025

Discovery of additional ancient genome duplications in yeasts

Whole genome duplication (WGD) has had profound macroevolutionary impacts on diverse lineages1,2, preceding adaptive radiations in vertebrates3-5, teleost fish6,7, and angiosperms8,9. In contrast to the many known ancient WGDs in animals10,11 and especially plants12-14, we are aware of evidence for only four in fungi15,16. The oldest of these occurred [~]100 million years ago (mya) and is shared by [~]60 extant Saccharomycetales species17,18, including the bakers yeast Saccharomyces cerevisiae (Fig. 1). Notably, this is the only known ancient WGD in the yeast subphylum Saccharomycotina. The dearth of ancient WGD events in fungi remains a mystery15. Some studies have suggested that fungal lineages that experience chromosome19 and genome15 duplication quickly go extinct, leaving no trace in the genomic record, while others contend that the lack of known WGD is due to an absence of data15,16. Under the second hypothesis, additional sampling and deeper sequencing of fungal genomes should lead to the discovery of more WGD events. Coupling hundreds of recently published genomes from nearly every described Saccharomycotina species with three additional long-read assemblies, we discovered three novel WGD events. While the functions of retained duplicate genes originating from these events are broad, they bear many similarities to the well-known WGD that occurred in the Saccharomycetales17. Our results suggest that WGD may be a more common evolutionary force in fungi than previously believed. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=111 SRC="FIGDIR/small/673279v1_fig1.gif" ALT="Figure 1"> View larger version (29K): org.highwire.dtl.DTLVardef@6a8f47org.highwire.dtl.DTLVardef@1e4a790org.highwire.dtl.DTLVardef@b4e0eforg.highwire.dtl.DTLVardef@11b7795_HPS_FORMAT_FIGEXP M_FIG O_FLOATNOFigure 1.C_FLOATNO Three additional whole genome duplications in Saccharomycotina yeasts. Rates of gene duplication are mapped onto Saccharomycetales (left) and Dipodascales (right) phylogenies. Whole genome duplications indicated along the lineages in which they are predicted to occur. Trees have been pruned for visualization purposes. C_FIG HighlightsO_LIEvidence for three whole genome duplications (WGDs) in Dipodascales yeasts C_LIO_LIImpacts of WGD are broad but bear many similarities to the known WGD in Saccharomycetales yeasts C_LIO_LIDuplicates with many protein-protein interactions are more likely to be retained over long timescales C_LIO_LIWGD in fungi is likely underreported C_LI

evolutionary biology↗

bioRxiv2025

Machine learning identifies novel signatures of antifungal drug resistance in Saccharomycotina yeasts

Antifungal drug resistance is a major challenge in fungal infection management. Numerous genomic changes are known to contribute to acquired drug resistance in clinical isolates of specific pathogens, but whether they broadly explain natural resistance across entire lineages is unknown. We leveraged genomic, ecological, and phenotypic trait data from naturally sampled strains from nearly all known species in subphylum Saccharomycotina to examine the evolution of resistance to eight antifungal drugs. The phylogenetic distribution of drug resistance varied by drug; fluconazole resistance was widespread, while 5-fluorocytosine resistance was rare, except in Lipomycetales. A random forest algorithm trained on genomic data predicted drug-resistant yeasts with 54-75% accuracy. In general, frequency of drug resistance correlated with prediction accuracy, with fluconazole resistance being consistently predicted with the highest accuracy (74.9%). Fluconazole resistance accuracy was similar between models trained on genome-wide variation in the presence and number of InterPro protein annotations across Saccharomycotina (74.9% accuracy) and those trained on amino acid sequence alignment data of Erg11, a protein known to be involved in fluconazole resistance (74.3-74.9% accuracy). Interestingly, the top Erg11 residues for predicting fluconazole resistance across Saccharomycotina do not overlap with, are not spatially close to, and are less conserved than those previously linked to resistance in clinical isolates of Candida albicans. In silico deep mutational scanning of the C. albicans Erg11 protein revealed that amino acid variants implicated in clinical cases of resistance are almost universally destabilizing while variants in our most informative residues are energetically more neutral, explaining why the latter are much more common than the former in natural populations. Importantly, previous experimental analyses of C. albicans Erg11 have shown that amino acid variation in our most informative residues, despite having never been directly implicated in clinical cases, can directly contribute to resistance. Our results suggest that studies of natural resistance in yeast species never encountered in the clinic will yield a fuller understanding of antifungal drug resistance.

microbiology↗

bioRxiv2025

An integrative genomic and chemical similarity approach linking fungal secondary metabolites and biosynthetic gene clusters

Fungi are well known to biosynthesize structurally complex secondary metabolites (SMs) with diverse bioactivities. These fungal SMs are frequently produced by biosynthetic gene clusters (BGCs). Linking SMs to their BGCs is key to understanding their chemical and biological functions. Reasoning that structural similarity of SMs arises from similarities in the genes involved in their biosynthesis, we developed an integrative approach that leverages known BGC-SM pairs to predict global links across SMs and BGCs in fungi. As proof of concept, we systematically interrogated metabolomes and genomes of 16 strains of the filamentous fungus Aspergillus fischeri, detecting a total of 60 metabolites. Of those, we were able to assign 22 to known BGC-SM pairs and propose specific hypotheses for the remaining 38 metabolites. These results suggest that coupling genomic similarity and chemical structure-based similarity is a straightforward and high-throughput approach for linking fungal SMs to their BGCs.

microbiology↗

bioRxiv2025

Ancient co-option of LTR retrotransposons as yeast centromeres

The evolutionary origins of the genetic point centromere in the brewers yeast Saccharomyces cerevisiae, a member of the order Saccharomycetales, are still unknown. Competing hypotheses suggest that the point centromere tripartite genetic centromere DNA elements (CDEs) either evolved from ancestral epigenetic centromeres by descent with modification or were gained through horizontal transfer from selfish DNA plasmids.1,2 Here, we identified centromeres in the sister order Saccharomycodales and termed them "proto-point centromeres" due to sequence features that bridge the evolutionary gap between point centromeres and ancestral centromeres types. Comparative genomic analyses across multiple yeast orders showed an unexpected evolutionary link between point and proto-point centromeres to the long terminal repeats (LTRs) of Ty5 retrotransposons. Strikingly, one Saccharomycodales species, Saccharomycodes ludwigii, harbors compact Ty5-based centromeres, where its CDEII elements are divergent AT-rich Ty5 LTRs. These living fossil centromeres show how retrotransposon cis-regulation was likely co-opted for genetic centromere specification. These insights show that point centromeres are direct descendants of retrotransposons and have evolved by descent with modification. Ultimately, the many diverse centromere types across the yeast subphylum may share a common ancestry rooted in retrotransposon activity.

evolutionary biology↗

bioRxiv2025

Stable hypermutators revealed by the genomic landscape of DNA repair genes among yeast species

Mutator phenotypes are short-lived due to the rapid accumulation of deleterious mutations. Yet, recent observations reveal that certain fungi can undergo prolonged accelerated evolution after losing DNA repair genes. Here, we surveyed 1,154 yeast genomes representing nearly all known yeast species of the subphylum Saccharomycotina to examine the relationship between reduced DNA repair repertoires and elevated evolutionary rates. We identified three distantly related lineages--encompassing 12% of species--with substantially reduced sets of DNA repair genes and the highest evolutionary rates in the entire subphylum. Two of these "faster-evolving lineages" (FELs)--a subclade within the order Pichiales and the Wickerhamiella/Starmerella (W/S) clade (order Dipodascales)--are described here for the first time, while the third corresponds to a previously documented Hanseniaspora FEL. Examination of DNA repair gene repertoires revealed a set of genes predominantly absent in these three FELs, suggesting a potential role in the observed acceleration of evolutionary rates. Genomic signatures in the W/S clade are consistent with a substantial mutational burden, including pronounced A|T bias and signatures of endogenous DNA damage. The W/S clade appears to mitigate UV-induced damage through horizontal acquisition of a bacterial photolyase gene, underscoring how gene loss may be offset by nonvertical evolution. These findings highlight how the loss of DNA repair genes gave rise to hypermutators that persist across macroevolutionary timescales, with horizontal gene transfer as an avenue for partial functional compensation.

evolutionary biology↗

bioRxiv2025

Genomic factors limiting the diversity of Saccharomycotina plant pathogens

The Saccharomycotina fungi have evolved to inhabit a vast diversity of habitats over their 400-million-year evolution. There are, however, only a few known fungal pathogens of plants in this subphylum, primarily belonging to the genera Eremothecium and Geotrichum. We compared the genomes of 12 plant-pathogenic Saccharomycotina strains to 360 plant-associated strains to identify features unique to the phytopathogens. Characterization of the oxylipin synthesis genes, a compound believed to be involved in Eremothecium pathogenicity, did not reveal any differences in gene presence within or between the plant-pathogenic and plant-associated strains. A reverse-ecological approach, however, revealed that plant pathogens lack several metabolic enzymes known to assist other phytopathogens in overcoming plant defenses. This includes L-rhamnose metabolism, formamidase and nitrilase genes. This result suggests that the Saccharomycotina plant pathogens are limited to infecting ripening fruits as they are without the necessary enzymes to degrade common phytohormones and secondary metabolites produced by plants.

genomics↗

bioRxiv2024

OrthoHMM: Improved Inference of Ortholog Groups using Hidden Markov Models

Accurate orthology inference is essential for comparative genomics and phylogenomics. However, orthology inference is challenged by sequence divergence, which is pronounced among anciently diverged organisms. We present OrthoHMM, an algorithm that infers orthologous gene groups using Hidden Markov Models parameterized from substitution matrices, which enables better detection of remote homologs. Benchmarking indicates OrthoHMM outperforms currently available methods; for example, using a curated set of Bilaterian orthogroups, OrthoHMM showed a 10.3 - 138.9% improvement in precision. Rank-based benchmarking using Bilaterian orthogroups and a novel dataset of orthogroups from organisms in three major eukaryotic kingdoms revealed OrthoHMM had the best overall performance (6.7 - 97.8% overall improvement). These findings suggest that Hidden Markov Models improve orthogroup inference.

bioinformatics↗

bioRxiv2024

Global patterns of species diversity and distribution in the biomedically and biotechnologically important fungal genus Aspergillus

Aspergillus fungi are key producers of pharmaceuticals, enzymes, and food products and exhibit diverse lifestyles, ranging from saprophytes to opportunistic pathogens. To improve understanding of Aspergillus species diversity, identify key environmental factors influencing their geographic distributions, and estimate the impact of future climate change, we trained a random forest machine learning classifier on 30,542 terrestrial occurrence records for 176 species ([~]40% of known species in the genus) and 96 environmental variables. We found that regions with high species diversity are concentrated in temperate forests, which suggests that areas with mild seasonal variation may serve as diversity hotspots. Species range estimates revealed extensive variability, both within and across taxonomic sections; while some species are cosmopolitan, others have more restricted ranges. Furthermore, range overlap between species is generally low. The top predictors of mean species richness were the index of cumulative human impact and five bioclimatic factors, such as temperature and temperate vs non-temperate ecoregions. Our future climate analyses revealed considerable variation in species range estimates in response to changing climates; some species ranges are predicted to expand (e.g., the food spoilage and mycotoxin-producing Aspergillus versicolor), and others are predicted to contract or remain stable. Notably, the predicted range of the major pathogen Aspergillus fumigatus was predicted to decrease in response to climate change, whereas the range of the major pathogen Aspergillus flavus was predicted to increase and gradually decrease. Our findings reveal how both natural and human factors influence Aspergillus species ranges and highlight their ecological diversity, including the diversity of their responses to changing climates, which is of relevance to pathogen and mycotoxin risk assessment.

ecology↗

bioRxiv2024

Gestation length drives brain size and litter size variation in eutherian mammals

The length of gestation in eutherian mammals, which is key to their reproductive success, is closely connected to other life history traits and with body mass and brain mass, but causal relationships between these variables are unclear. Here, we used an integrative analytical framework to evaluate the evolutionary relationships between gestation length and eight other traits on a dataset of 3,258 eutherian mammals and infer causality. We identify variation in generation length and litter size as the primary predictors of eutherian gestation length variation, whereas additional traits, such as brain mass, significantly predict gestation length only in specific mammalian orders. Using a structural equation modeling approach known as phylogenetic path analysis to infer causality, we find that gestation length variation positively influences brain mass variation and negatively influences litter size variation. Furthermore, body mass causally influences gestation length variation only in certain orders. Consistent with these causal inferences, examination of trait-trait coevolution reinforces that gestation length is strongly positively associated with brain mass, strongly negatively associated with litter size, and only moderately correlated with body mass. These findings reveal how gestation length directly and indirectly influences, and is influenced by, other key eutherian traits. Our study establishes a robust framework for identifying causal relationships within suites of correlated and co-evolving traits.

evolutionary biology↗

bioRxiv2024

Machine learning uncovers gene families impacting oxidative stress resistance across yeasts

Reactive oxygen species (ROS) are highly reactive molecules encountered by yeasts during routine metabolism and during interactions with other organisms, including host infection. Here, we characterized the variation in resistance to ROS across the ancient yeast subphylum Saccharomycotina and used machine learning (ML) to identify gene families whose sizes were predictive of ROS resistance. The most predictive features were enriched in gene families related to cell wall organization and included two reductase gene families. We estimated the quantitative contributions of features to each species classification to guide experimental validation and showed that overexpression of the old yellow enzyme (OYE) reductase increased ROS resistance in Kluyveromyces lactis, while Saccharomyces cerevisiae mutants lacking multiple mannosyltransferase-encoding genes were hypersensitive to ROS. Altogether, this work provides a framework for how ML can uncover genetic mechanisms underlying trait variation across diverse species and inform trait manipulation for clinical and biotechnological applications.

microbiology↗