Biology subjects

Colome-Tatche, M.

Publications and source records attributed to Colome-Tatche, M..

At least 19 recordsLinked to original sources

bioRxiv2026

XplainTDA: Explaining the topological structure of single-cell multi-omics data through perturbation

Single-cell experiments have improved our ability to study biological systems by profiling thousands of cells simultaneously. The data can be viewed as a point cloud in a proper experimental space, whose shape is defined by the gene regulatory network. Persistent homology has been proposed as a method for studying the intrinsic structure of data and has shown the potential to reveal novel insights. Existing approaches focus on detecting topological features or groups of cells, without a focus on explainability and interpretability. Here, we introduce an explainable topological data analysis framework for single-cell data (XplainTDA) that identifies molecular determinants underlying the topological structure of the data via a perturbation-based approach. Across omics and data sets, we show that the method is stable, reproducible, and interpretable, extending persistent homology from topological description to molecular interpretation.

bioinformatics↗

bioRxiv2026

BATTLE-AMP: Benchmarking Antimicrobial Peptide Predictors

As antimicrobial resistance outpaces antibiotic development, antimicrobial peptides (AMPs) have emerged as a promising class of alternative antibacterials, and computational predictors are increasingly used to prioritize AMP candidates. Such predictors are typically evaluated on binary AMP/non-AMP classification, which does not test whether they can identify peptides with clinically relevant potency against specific pathogens. We present BATTLE-AMP, a benchmarking framework that evaluates AMP predictors against experimentally measured minimum inhibitory concentrations (MICs) across clinically relevant bacterial species and strains. We surveyed 48 published methods, finding fewer than 25% reproducible, and benchmarked 10 model families (21 variants) using experimental MIC data, synthetic sequence perturbations, activity cliff analyses, and all-atom molecular dynamics (MD) simulations. Four findings emerge: (i) models trained on MIC data outperform binary classifiers regardless of architecture; (ii) the best model depends on the target pathogen, so model selection must be guided by the biological question; (iii) most models cannot distinguish active peptides from inactive sequences with identical amino acid composition; and (iv) activity cliffs remain unresolved by both machine learning and MD, marking a limit of current computational methods. BATTLE-AMP is released as an open Snakemake framework at https://github.com/szczurek-lab/battleamp-snakemake for benchmarking new models and scoring novel candidate libraries.

bioinformatics↗

bioRxiv2026

A transient CD87-centred axis enhances the Th17 potential of cDC2 in neonates

Type 2 conventional dendritic cells (cDC2) orchestrate T cell immunity, yet the signals that regulate their tissue-specific functions across development remain poorly defined. We demonstrate that splenic cDC2/DC3 subset heterogeneity is established perinatally and that splenic ESAMhi cDC2A undergo a transcriptional and functional transition around the time of weaning, which occurs independently of microbiota. Comparative transcriptomics identified neonatal-specific regulatory programs, including elevated expression of Plaur, encoding CD87/uPAR. CD87 sensitizes neonatal ESAMhi cDC2A to coagulation factor XII (FXII), thereby enhancing their capacity to promote Th17 differentiation. In human infants, CD87 expression was high on cDC2, its expression declined with age and in plasma of preterm infants CD87 positively correlated with Th17-associated cytokines. Together, we identify a conserved link between coagulation pathways and cDC2 developmental programming that may offer new opportunities to modulate Th17 responses in infancy. SummarySplenic cDC2 undergo a transcriptional and functional remodeling around weaning. Elevated CD87 expression on neonatal cDC2 enhances their FXII-driven Th17 potential. Conserved in human infants, elevated CD87 expression reveals a coagulation-linked pathway shaping early-life dendritic cell function and suggests targets to improve vaccination.

immunology↗

bioRxiv2026

Periweaning diet-induced activation of an IFNγ-mediated regulatory circuit promotes the homeostasis of cytotoxic CD8+ T Cells

Balancing pathogen defence with maintaining tolerance to environmental antigens, such as food or commensals, in neonates is essential for survival and the establishment of life-long immune homeostasis. Instructed by environmental signals type 1 conventional dendritic cells (cDC1) drive either T cell tolerance or immunity. Here, we uncover an interferon (IFN)-{gamma}-driven regulatory circuit in early life that relays dietary cues to spleen cDC1. Loss-of-function demonstrates that IFN{gamma}-mediated STAT1-signaling induces an immunogenic maturation program in spleen cDC1 that instructs cDC1 to expand effector memory CD8 T cells. This program emerges during weaning, when IFN{gamma} production from lymphocytes rises, it occurs in germ-free mice and remains responsive to dietary intervention in adult mice. During the transition from breastfeeding to solid food at weaning, this circuit relays dietary information to spleen cDC1 to shape the effector phenotype of food-antigen specific CD8+ T cells in a feed-forward manner, allowing cDC1 to recalibrate the T cell pool at the moment of nutritional independence.

immunology↗

bioRxiv2026

Centromeres are hotspots of cytosine methylation epimutations in a filamentous fungus

Epimutations are changes in chromatin modifications, such as DNA methylation or histone modifications. Some of these epigenetic changes can be inherited for several generations, and they potentially contribute to evolutionary processes. Estimates of epimutation rates now exists in a few species, but the presence and function of epigenetic marks are not conserved across different species. To understand the properties of epimutations in fungi, we performed a mutation accumulation experiment with the filamentous fungus Neurospora crassa and investigated spontaneous changes in DNA methylation and trimethylation of lysine 9 on histone H3 (H3K9me3) in the mutation accumulation lines. We observed that centromeric regions are hotspots of spontaneous DNA methylation changes in N. crassa. In these hotspot regions, DNA methylation changes were transmitted across mitoses, but changes occurring in euchromatin were not maintained. The rate of DNA methylation changes was around 30 000 fold faster than the genetic mutation rate. We did not observe spontaneous changes in H3K9me3 that were transmitted across mitoses. Our results show that while spontaneous epimutations occur in this species, they occur predominantly in gene poor heterochromatic regions, so their impact for evolutionary adaptation may be limited.

genetics↗

bioRxiv2026

Integration of single-cell multi-omic data with graph-based topic modelling

Recent advances in single-cell biology enable the profiling of multiple molecular layers, such as the transcriptome, epigenome, and surface proteins, within a single cell. Tackling the complexity of these data from different perspectives allows researchers to get the most complete insights into the biological properties of cells. Here, we propose a graph-based topic modelling method called bionSBM. Our method leverages well-known community-detection methods for multipartite graphs and the interpretability of topic modelling to cluster and explain high-dimensional, sparse, and noisy single-cell matrices. We applied our algorithm to paired single-cell multi-omics data, such as 10X Multiome, SHARE-seq, and CITE-seq. We showed that it achieves superior performance compared to state-of-the-art methods for ground-truth label retrieval, with high specificity and distinct biological interpretability.

bioinformatics↗

bioRxiv2025

Geometry aware graph attention networks to explain single-cell chromatin state and gene expression

High-throughput measurements that profile the transcriptome or the epigenome of single-cells are becoming a common way to study cell identity. These data are high dimensional, sparse and non linear. Here we present SEAGALL (Single-cell Explainable Geometry-Aware Graph Attention Learning pipeLine), a hypothesis free method to extract biologically relevant features from single-cell experiments based on geometry regularised autoencoders (GRAE) and explainable graph attention networks (GAT). We use a GRAE to embed the data into a latent space preserving the data geometry and we construct a cell-to-cell graph computing distances in the GRAE bottleneck. Exploiting the attention mechanism to dynamically learn the relevant edges, we use GATs to classify the cells and we explain the predictions of the model with XAI methods to unravel the features which are driving cell identity beyond marker genes. We apply our method to data sets from scRNA-seq, scATAC-seq and scChIP-seq experiments. SEAGALL can extract cell type specific and stable signatures which not only differ from the ones found in classical linear approaches but are less biassed by coverage and high expression.

bioinformatics↗

bioRxiv2024

Benchmarking scRNA-seq copy number variation callers

Copy number variations (CNVs), the gain or loss of genomic regions, are associated with different diseases and cancer types, where they are related to tumor progression and treatment outcome. Single cell technologies offer new possibilities to measure CNVs in individual cells, allowing to assess population heterogeneity and to delineate subclonal structures. Single cell whole-genome sequencing is considered the gold-standard for the quantification of CNVs in single cells. However, the majority of existing single cell datasets interrogate gene expression, using scRNA-seq. Consequently, several computational approaches have been developed to identify CNVs from that data modality. Nevertheless, an independent benchmarking of these methods is lacking. We used 15 scRNA-seq datasets and evaluated six popular computational methods in their ability to recover the ground truth CNVs using a large set of performance metrics. Additionally, we explored whether they could correctly identify euploid cells, especially also in fully diploid samples, and subclonal structures in heterogeneous tumor samples. We discovered several dataset-specific factors that influence the performance of the methods, such as the dataset size and the number and type of CNVs in the analyzed sample. We found that the choice of the reference dataset can have a large impact on the performance. Methods which included additional allelic information from the scRNA-seq reads performed more robustly across scenarios, but at the cost of higher runtime. Furthermore, the methods differed substantially in their additional functionalities and resource requirements. We offer a benchmarking pipeline to help identify the optimal CNV calling method for newly generated scRNA-seq datasets, and to benchmark and improve new methods performance.

bioinformatics↗

bioRxiv2024

Path-based reasoning in biomedical knowledge graphs

Understanding complex interactions in biomedical networks is crucial for advancements in biomedicine, but traditional link prediction (LP) methods are limited in capturing this complexity. Representation-based learning techniques improve prediction accuracy by mapping nodes to low-dimensional embeddings, yet they often struggle with interpretability and scalability. We present BioPathNet, a novel graph neural network framework based on the Neural Bellman-Ford Network (NBFNet), addressing these limitations through path-based reasoning for LP in biomedical knowledge graphs. Unlike node-embedding frameworks, BioPathNet learns representations between node pairs by considering all relations along paths, enhancing prediction accuracy and interpretability. This allows visualization of influential paths and facilitates biological validation. BioPathNet leverages a background regulatory graph (BRG) for enhanced message passing and uses stringent negative sampling to improve precision. In evaluations across various LP tasks, such as gene function annotation, drug-disease indication, synthetic lethality, and lncRNA-mRNA interaction prediction, BioPathNet consistently outperformed shallow node embedding methods, relational graph neural networks and task-specific state-of-the-art methods, demonstrating robust performance and versatility. Our study predicts novel drug indications for diseases like acute lymphoblastic leukemia (ALL) and Alzheimers, validated by medical experts and clinical trials. We also identified new synthetic lethality gene pairs and regulatory interactions involving lncRNAs and target genes, confirmed through literature reviews. BioPathNets interpretability will enable researchers to trace prediction paths and gain molecular insights, making it a valuable tool for drug discovery, personalized medicine and biology in general.

bioinformatics↗

bioRxiv2024

Cross-species analyses reveal RORγt-expressing dendritic cells are a lineage of antigen presenting cells conserved across tissues

Conventional dendritic cells (cDCs) are potent antigen presenting cells (APCs) that exhibit tissue and age-specific diversity allowing them to direct situation-adapted immunity. Thereby they harbor great potential for being targeted in vaccination and cancer. Here, we resolve conflicting data about expression of retinoic acid receptor-related orphan receptor-{gamma}t (ROR{psi}t) in cDCs. We show that ROR{psi}t+ DCs exist in murine lymphoid and non-lymphoid tissues across age. Fate mapping, functional assays and single cell multiomic profiling reveal these cells as ontogenetically and transcriptionally distinct from other well characterized cDC subtypes, as well as from ROR{psi}t+ type 3 innate lymphocytes (ILC3s). We show that ROR{psi}t+ DCs can migrate to lymph nodes and activate naive CD4+ T cells in response to inflammatory triggers. Comparative and cross-species transcriptomics revealed homologous populations in human spleen, lymph nodes and intestines. Further, integrated meta-analyses aligned ROR{psi}t+ DCs identified here with other emerging populations of ROR{psi}t+APCs, including R-DC-like cells, Janus cells/extrathymic Aire expressing cells (eTACs) and subtypes of Thetis cells. While ROR{psi}t+APCs have primarily been linked to T cell tolerance, our work establishes ROR{psi}t+ DCs as unique lineage of immune sentinel cells conserved across tissues and species that expands the functional repertoire of ROR{psi}t+ APCs beyond promoting tolerance. One sentence summaryROR{gamma}t+ DC exhibit versatile APC functions and are a distinct immune lineage conserved across age, tissues and species that entails Thetis cells, Janus cells/ROR{gamma}t+ eTACs and R-DC-like cells.

immunology↗

bioRxiv2024

High-resolution scRNA-seq reveals genomic determinants of antigen expression hierarchy in African Trypanosomes

Antigenic variation is an immune evasion strategy used by many different pathogens. It involves the periodic, non-random switch in the expression of different antigens throughout an infection. How the observed hierarchy in antigen expression is achieved has remained a mystery. A key challenge in uncovering this process has been the inability to track transcriptome changes and potential genomic rearrangements in individual cells during a switch event. Here, we report the establishment of a highly sensitive single-cell RNA-seq (scRNA-seq) approach for the model protozoan parasite Trypanosoma brucei. This approach has revealed genomic rearrangements that occur in individual cells during a switch event. Our data show that following a double-strand break (DSB) in the transcribed antigen-coding gene - an important trigger for antigen switching - the type of repair mechanism and the resultant antigen expression depend on the availability of a homologous repair template in the genome. When such a template was available, repair proceeded through segmental gene conversion, creating new, mosaic antigen-coding genes. Conversely, in the absence of a suitable template, a telomere-adjacent antigen-coding gene from a different part of the genome was activated by break-induced replication. Our results reveal the critical role of available repair sequence in the antigen selection mechanism. Additionally, our study demonstrates the power of highly sensitive scRNA-seq methods in detecting genomic rearrangements that drive transcriptional changes at the single-cell level.

microbiology↗

bioRxiv2024

Community assesses differential cell communication using large multi-sample case-control scRNAseq datasets.

Cell-cell communication is essential for physiological tissue function. In disease, this communication often gets disbalances by changes in the tissue cell type composition, fraction of cell engaged in communication and the rising or dropping expression levels of ligands, receptors and adhesion molecules. The changes in all these components of communication can be studied using single cell RNA-sequencing (scRNAseq) methods. With dropping sequencing costs, it is now possible to perform scRNAseq studies in larger cohorts of case and control samples to better address the heterogeneity of diseases. Here we present community, an R-based tool that is designed to perform differential communication analysis using scRNAseq between large cohorts of cases and controls. Community is able to reconstruct communication between different cell types both in the case and the control cohort of a dataset, and subsequently analyze which communication channels are affected in disease. Community is the first tool that integrates cell type abundance into the calculation of an interaction strength. Community is also able to disentangle the mechanisms underlying these changes, as well as detect interactions that are kept compensated by a sender or a receiver despite the disbalanced signaling from the counterpart. We tested community on two disease entities, ulcerative colitis and acute myeloid leukemia, using published scRNAseq datasets. We compared the performance of our tool to other differential communication pipelines, which community outperformed in speed and robustness. Overall, community is a fast, well-scalable, user-friendly R tool to assess differential cell-cell communication using large case-control scRNAseq datasets disentangling the driving mechanisms of communication shifts in disease.

bioinformatics↗

bioRxiv2023

A compendium of Amplification-Related Gain Of Sensitivity (ARGOS) genes in human cancer

Chromosomal gains are among the most frequent somatic genetic alterations occurring in cancer. While the effect of sustained oncogene expression has been characterized, the impact of copy-number gains affecting collaterally-amplified "bystander" genes on cellular fitness remains less understood. To investigate this, we built a comprehensive map of dosage compensations across human cancers by integrating expression and copy number profiles from over 8,000 TCGA tumors and CCLE cell lines. Further, we analyzed the effect of gene overexpression across 17 human cancer ORF screens to provide an overview of genes that prove toxic to cancer cells when overexpressed. Combining these two independent approaches we propose a class of Amplification-Related Gain Of Sensitivity (ARGOS) genes. These genes are located in commonly amplified regions of the genome, have lower expression levels than expected by their copy-number status, and are toxic to cancer cells when overexpressed. We experimentally validated CDKN1A and RBM14 as high-confidence pan-cancer ARGOS genes in lung and breast cancer cell line models. We additionally suggest that RBM14s mechanism of toxicity involves altered DNA damage response and innate immune signaling processes following gene overexpression. Finally, we provide a comprehensive catalog of compensated, toxic, and ARGOS genes as a community resource.

cancer biology↗

bioRxiv2023

ARID1A Mutations Protect Follicular Lymphoma from FAS-dependent Immune Surveillance by Reducing RUNX3/ETS1-Driven FAS-Expression

The cell death receptor FAS and its ligand (FASLG) play crucial roles in the selection of B cells during the germinal center (GC) reaction. Failure to eliminate potentially harmful B cells via FAS can lead to lymphoproliferation and the development B cell malignancies. The classic form of follicular lymphoma (FL) is a prototypic GC-derived B cell malignancy, characterized by the t(14;18) (q32;q21)IGH::BCL2 translocation and overexpression of antiapoptotic BCL2. Additional alterations were shown to be clinically relevant, including mutations in ARID1A. ARID1A is part of the SWI/SNF nucleosome remodeling complex that regulates DNA accessibility ("openness"). However, the mechanism how ARID1A mutations contribute to FL pathogenesis remains unclear. We analyzed 151 FL biopsies of patients with advanced stage disease at initial diagnosis and found that ARID1A mutations were recurrent and mainly disruptive, with an overall frequency of 18%. Additionally, we observed that ARID1A mutant FL showed significantly lower FAS protein expression in the FL tumor cell population. Functional experiments in BCL2-translocated lymphoma cells demonstrated that ARID1A is directly involved in the regulation of FAS, and ARID1A loss leads to decreased FAS protein and gene expression. However, ARID1A loss did not affect FAS promotor openness. Instead, we identified and experimentally validated a previously unknown co-transcriptional complex consisting of RUNX3 and ETS1 that regulates FAS expression, and ARID1A loss leads to reduced RUNX3 promotor openness and gene expression. The reduced FAS levels induced by ARID1A loss rendered lymphoma cells resistant to both soluble and T cell membrane-anchored FASLG-induced apoptosis. In summary, we have identified a functionally and clinically relevant mechanism how FL cells can escape FAS-dependent immune surveillance, which may also impact the efficacy of T cell-based therapies, including bispecific antibodies and CAR T cells.

cancer biology↗

bioRxiv2022

Single-cell metabolic profiling reveals subgroups of primary human hepatocytes showing heterogeneous responses to drug challenge.

Xenobiotics are primarily metabolized by hepatocytes in the liver, and primary human hepatocytes (PHHs) are the gold standard model for the assessment of drug efficacy, safety and toxicity in the early phases of drug development. Recent advances in single-cell genomics have shown liver zonation and ploidy as main drivers of cellular heterogeneity. However, little is known about the impact of hepatocyte specialization on liver function upon metabolic challenge, including hepatic metabolism, detoxification, and protein synthesis. Here, we investigate the metabolic capacity of individual human hepatocytes in vitro, and assess how chronic accumulation of lipids enhances cellular heterogeneity and impairs the metabolisms of drugs. A phenotyping five-probe cocktail was used to identify four functional subgroups of hepatocytes that respond differently to drug challenge and fatty acid accumulation. These four subgroups display differential gene expression profiles upon cocktail treatment and xenobiotic metabolism-related specialization. Notably, intracellular fat accumulation leads to increased transcriptional variability and diminished the drug-related metabolic capacity of hepatocytes. Our results demonstrate that, upon a metabolic challenge such as exposure to drugs or intracellular fat accumulation, hepatocyte subgroups lead to different and heterogeneous transcriptional responses.

cell biology↗

bioRxiv2022

epiAneufinder: identifying copy number variations from single-cell ATAC-seq data

Single-cell open chromatin profiling via the single-cell Assay for Transposase-Accessible Chromatin using sequencing (scATAC-seq) assay has become a mainstream measurement of open chromatin in single-cells. Here we present a novel algorithm, epiAneufinder, that exploits the read count information from scATAC-seq data to extract genome-wide copy number variations (CNVs) for individual cells, allowing to explore the CNV heterogeneity present in a sample at the single-cell level. Using different cancer scATAC-seq datasets, we show how epiAneufinder can identify intratumor clonal heterogeneity in populations of single cells based on their CNV profiles. These profiles are concordant with the ones inferred from single-cell whole genome sequencing data for the same samples. epiAneufinder allows the addition of single-cell CNV information to scATAC-seq data, without the need of additional experiments, unlocking a layer of genomic variation which is otherwise unexplored.

bioinformatics↗

bioRxiv2020

Region-level Epimutation Rates in Arabidopsis thaliana

Failure to maintain DNA methylation patterns during plant development can occasionally give rise to so-called spontaneous epimutations. These stochastic methylation changes are sometimes heritable across generations and thus accumulate in plant genomes over time. Recent evidence indicates that spontaneous epimutations have a major role in shaping patterns of methylation diversity in plant populations. Using single CG dinucleotides as units of analysis, previous work has shown that the epimutation rate is several orders of magnitude higher than the genetic mutation rate. While these large rate differences have obvious implications for understanding genome-methylome co-evolution, the functional relevance of single CG methylation changes remains questionable. In contrast to single CG, solid experimental evidence has linked methylation gains and losses in larger genomic regions with transcriptional variation and heritable phentoypic effects. Here we show that such region-level changes arise stochastically at about the same rate as those at individual CG sites, are only marginal dependent on region size and cytosine density, but strongly dependent on chromosomal location. We also find consistent evidence that region-level epimutations are not restricted to CG contexts but also frequently occur in non-CG regions at the genome-wide scale. Taken together, our results support the view that many differentially methylated regions (DMRs) in natural populations originate from epimutational events and may not be effectively tagged by proximal SNPs. This possibility reinforces the need for epigenome-wide association studies (EWAS) in plants as away to identify the epigenetic basis of adaptive traits.

plant biology↗

bioRxiv2020

Benchmarking atlas-level data integration in single-cell genomics

Cell atlases often include samples that span locations, labs, and conditions, leading to complex, nested batch effects in data. Thus, joint analysis of atlas datasets requires reliable data integration. Choosing a data integration method is a challenge due to the difficulty of defining integration success. Here, we benchmark 38 method and preprocessing combinations on 77 batches of gene expression, chromatin accessibility, and simulation data from 23 publications, altogether representing >1.2 million cells distributed in nine atlas-level integration tasks. Our integration tasks span several common sources of variation such as individuals, species, and experimental labs. We evaluate methods according to scalability, usability, and their ability to remove batch effects while retaining biological variation. Using 14 evaluation metrics, we find that highly variable gene selection improves the performance of data integration methods, whereas scaling pushes methods to prioritize batch removal over conservation of biological variation. Overall, BBKNN, Scanorama, and scVI perform well, particularly on complex integration tasks; Seurat v3 performs well on simpler tasks with distinct biological signals; and methods that prioritize batch removal perform best for ATAC-seq data integration. Our freely available reproducible python module can be used to identify optimal data integration methods for new data, benchmark new methods, and improve method development.

bioinformatics↗