Biology subjects

Dufault-Thompson, K.

Publications and source records attributed to Dufault-Thompson, K..

7 recordsLinked to original sources

bioRxiv2026

LAMBDA: A Prophage Detection Benchmark for Genomic Language Models

Transformer-based genomic sequence models represent an emerging frontier in computational biology. Yet, their embeddings have not yet shown the same level of predictive power as natural and protein language models, highlighting a gap between current implementations and theoretical promise. Existing benchmarks for DNA language models primarily focus on classifying regulatory elements in eukaryotic genomes, leaving open the fundamental question of whether these models learn sequence-level features across whole genomes. We introduce LAMBDA, a benchmark designed to rigorously evaluate genome language model embeddings through phage-bacteria sequence discrimination across four categories of increasing complexity: probing tasks, fine-tuning assessments, diagnostic tests, and genome-wide prophage detection. Our comprehensive analysis of current genomic language models provides insight into the importance of training data selection relative to model size, the need for domain-specific training, and the capabilities and limitations of genomic language models for detecting prophage sequences. This benchmark represents a challenging genomic annotation task in the bacterial domain and addresses a key computational problem with direct relevance to microbiology and medicine.

genomics↗

bioRxiv2025

Ancestral proteins trace the emergence of substrate specificity and oligomerization within bacterial DEDDy dinucleases

Nucleases play a crucial role in bacterial physiology, influencing processes such as DNA repair, genome maintenance, and host-pathogen interactions. We recently identified a class of nucleases, diDNases, which are encoded on mobile genetic elements and homologous to the house-keeping nanoRNase C (NrnC). Despite their shared structural fold, diDNases and NrnC orthologs exhibit differences. DiDNases form dimers and preferably cleave DNA dinucleotides, whereas NrnC homologs assemble into octamers that do not discriminate between RNA or DNA dinucleotides. Here, we investigate the evolutionary divergence of these enzymes using ancestral sequence reconstruction. Our results show that both diDNases and NrnC orthologs originated from a dimeric ancestor with intermediate substrate preferences. Structural analyses of ancestral and extant dinucleases provide a molecular rational for how gradual changes in conformation gave rise to substrate preferences, oligomeric state, and catalytic efficiency of these related, yet distinct enzyme clades. These findings provide insights into how small structural modifications enable large-scale changes in molecular assembly and functional specialization harnessing a conserved protein fold. In addition, the preference of the early ancestors for DNA dinucleotides and preservation of this activity in all extant enzymes strongly argues for a biological function of DNA dinucleotides.

evolutionary biology↗

bioRxiv2024

Gut Bacteria Metabolize Natural and Synthetic Steroid Hormones via the Reductive OsrABC Pathway

Steroid hormone metabolism by the gut microbiome affects host physiology, but the underlying microbial pathways remain incompletely understood. Here, we isolate a novel human gut bacterium, Clostridium steroidoreducensT strain HCS.1 that reduces cortisol and related steroid hormones to 3{beta},5{beta}-tetrahydrosteroid products. Through transcriptomics and enzymatic discovery, we establish the C. steroidoreducens OsrABC steroid hormone pathway. OsrA is a 3-oxo-{Delta}1-steroid hormone reductase that targets synthetic glucocorticoids, including prednisolone-- a frontline Crohns disease therapy. OsrB is a 3-oxo-{Delta}4-steroid reductase that converts steroid hormones to 5{beta}-dihydrosteroid intermediates, which OsrC subsequently reduces to 3{beta},5{beta}-tetrahydro products. Homologs of osrA and osrB predict steroid-reducing activity across gut bacteria and are enriched in Crohns disease patient metagenomes. Consistent with a role in modulating drug efficacy, C. steroidoreducens colonization decreases prednisolone bioavailability in gnotobiotic mice. These findings thus define a previously unrecognized pathway for microbial steroid hormone inactivation and establish a mechanistic basis for bacterial interference with anti-inflammatory therapies.

microbiology↗

bioRxiv2023

Deciphering Phage-Host Specificity Based on the Association of Phage Depolymerases and Bacterial Surface Glycan with Deep Learning

Phage tailspike proteins are depolymerases that target diverse bacterial surface glycans with high specificity, determining the host-specificity of numerous phages. To address the challenge of identifying tailspike proteins due to their sequence diversity, we developed SpikeHunter, an approach based on the ESM-2 protein language model. Using SpikeHunter, we successfully identified 231,965 tailspike proteins from a dataset comprising 8,434,494 prophages found within 165,365 genomes of five common pathogens. Among these proteins, 143,035 tailspike proteins displayed strong associations with serotypes. Moreover, we observed highly similar tailspike proteins in species that share closely related serotypes. We found extensive domain swapping in all five species, with the C-terminal domain being significantly associated with host serotype highlighting its role in host range determination. Our study presents a comprehensive cross-species analysis of tailspike protein to serotype associations, providing insights applicable to phage therapy and biotechnology.

bioinformatics↗

bioRxiv2022

Reconstruction and analysis of thermodynamically-constrained models reveal metabolic responses of a deep-sea bacterium to temperature perturbations

Microbial acclimation to different temperature conditions can involve broad changes in cell composition and metabolic efficiency. A systems-level view of these metabolic responses in non-mesophilic organisms, however, is currently missing. In this study, thermodynamically-constrained genome-scale models were applied to simulate the metabolic responses of a deep-sea psychrophilic bacterium, Shewanella psychrophila WP2, under suboptimal (4{degrees}C), optimal (15{degrees}C), and supraoptimal (20{degrees}C) growth temperatures. The models were calibrated with experimentally determined growth rates of WP2. Gibbs free energy change of reactions ({Delta}rG), metabolic fluxes, and metabolite concentrations were predicted using random simulations to characterize temperature-dependent changes in the metabolism. The modeling revealed the highest metabolic efficiency at the optimal temperature, and it suggested distinct patterns of ATP production and consumption that could lead to the lower metabolic efficiency under suboptimal or supraoptimal temperatures. The modeling also predicted rearrangement of fluxes through multiple metabolic pathways, including the glycolysis pathway, Entner-Doudoroff pathway, tricarboxylic acid (TCA) cycle, and the electron transport system, and these predictions were corroborated through comparisons to WP2 transcriptomes. Furthermore, predictions of metabolite concentrations revealed the potential conservation of reducing equivalents and ATP in the suboptimal temperature, consistent with experimental observations from other psychrophiles. Taken together, the WP2 models provided mechanistic insights into the metabolism of a psychrophile in response to different temperatures. ImportanceMetabolic flexibility is a central component of any organisms ability to survive and adapt to changes in environmental conditions. This study represents the first application of thermodynamically-constrained genome-scale models in simulating the metabolic responses of a deep-sea psychrophilic bacterium to varying temperatures. The models predicted differences in metabolic efficiency that were attributed to changes in metabolic pathway utilization and metabolite concentration during growth under optimal and non-optimal temperatures. Experimental growth measurements were used for model calibration, and temperature-dependent transcriptomic changes corroborated the model-predicted rearrangement of metabolic fluxes. Overall, this study highlights the utility of modeling approaches in studying the temperature-driven metabolic responses of an extremophilic organism.

systems biology↗

bioRxiv2022

Putative host-derived insertions in the genome of circulating SARS-CoV-2 variants

Insertions in the SARS-CoV-2 genome have the potential to drive viral evolution, but the source of the insertions is often unknown. Recent proposals have suggested that human RNAs could be a source of some insertions, but the small size of many insertions makes this difficult to confirm. Through an analysis of available direct RNA sequencing data from SARS-CoV-2 infected cells, we show that viral-host chimeric RNAs are formed through what are likely stochastic RNA-dependent RNA polymerase template switching events. Through an analysis of the publicly available GISAID SARS-CoV-2 genome collection, we identified two genomic insertions in circulating SARS-CoV-2 variants that are identical to regions of the human 18S and 28S rRNAs. These results provide direct evidence of the formation of viral-host chimeric sequences and the integration of host genetic material into the SARS-CoV-2 genome, highlighting the potential importance of host-derived insertions in viral evolution. IMPORTANCEThroughout the COVID-19 pandemic, the sequencing of SARS-CoV-2 genomes has revealed the presence of insertions in multiple globally circulating lineages of SARS-CoV-2, including the Omicron variant. The human genome has been suggested to be the source of some of the larger insertions, but evidence for this kind of event occurring is still lacking. Here, we leverage direct RNA sequencing data and SARS-CoV-2 genomes to show host-viral chimeric RNAs are generated in infected cells and two large genomic insertions have likely been formed through the incorporation of host rRNA fragments into the SARS-CoV-2 genome. These host-derived insertions may increase the genetic diversity of SARS-CoV-2 and expand its strategies to acquire genetic materials, potentially enhancing its adaptability, virulence, and spread.

bioinformatics↗

bioRxiv2020

Diversity and Function of the Eastern Oyster (Crassostrea virginica) Microbiome

Marine invertebrate microbiomes play important roles in various host and ecological processes. However, a mechanistic understanding of host-microbe interactions is so far only available for a handful of model organisms. Here, an integrated taxonomic and functional analysis of the microbiome of the eastern oyster, Crassostrea virginica, was performed using 16S rRNA gene amplicon profiling, shotgun metagenomics, and genome-scale metabolic reconstruction. A relatively low number of amplicon sequence variants (ASVs) were observed in oyster tissues compared to water samples, while high variability was observed across individual oysters and among different tissue types. Targeted metagenomic sequencing of the gut microbiota led to further characterization of a dominant bacterial taxon, the class Mollicutes, which was captured by the reconstruction of a metagenome-assembled genome (MAG). Genome-scale metabolic reconstruction of the oyster Mollicutes MAG revealed a reduced set of metabolic functions and a high reliance on the uptake of host-derived nutrients. A chitin degradation and an arginine deiminase pathway were unique to the MAG as compared to other closely related Mycoplasma genomes, indicating a distinct mechanism of carbon and energy acquisition by the oyster- associated Mollicutes. A systematic reanalysis of public eastern oyster-derived microbiome data revealed the Mollicutes as a ubiquitous taxon among adult oysters despite their general absence in larvae and biodeposit samples, suggesting potential horizontal transmission via an unknown mechanism. IMPORTANCEDespite well-documented biological significance of invertebrate microbiomes, a detailed taxonomic and functional characterization is frequently missing from many non-model marine invertebrates. By using 16S rRNA gene-based community profiling, shotgun metagenomics, and genome-scale metabolic reconstruction, this study provides an integrated taxonomic and functional analysis of the microbiome of the eastern oyster, Crassostrea virginica. Community profiling revealed a surprisingly low richness, as compared to surrounding seawater, and high variability among different tissue types and individuals. Reconstruction of a Mollicutes MAG enabled the phylogenomic positioning and functional characterization of the oyster-associated Mollicutes. Comparative analysis of the adult oyster gut, biodeposits, and oyster larvae samples indicated the potentially ubiquitous associations of the Mollicutes taxon with adult oysters. To the best of our knowledge, this study represented the first metagenomics derived functional inference of the eastern oyster microbiome. An integrated analytical procedure was developed for the functional characterization of microbiomes in other non-model host species.

bioinformatics↗