Biology subjects

Grover, A.

Publications and source records attributed to Grover, A..

13 recordsLinked to original sources

bioRxiv2026

Chiron3D: an interpretable deep learning framework for understanding the DNA code of chromatin looping

MotivationThree-dimensional folding of the genome into structures such as chromatin loops is essential for gene regulation. Current experimental methods for mapping these structures, like Hi-C and HiChIP, are labor-intensive and require repeated assays to test hypothesized mutation effects. This motivates the need for predictive approaches that reveal the sequence determinants of chromatin loops. ResultsIn this work, we present a novel and interpretable computational pipeline for predicting CTCF-mediated chromatin loops. We propose Chiron3D, a DNA-only model trained in a cell-type specific manner to predict CTCF HiChIP contact maps. By leveraging pre-trained embeddings from a foundation model, our approach is competitive with baselines that take CTCF ChIP-seq as additional input, while enabling nucleotide-level attribution to the input DNA sequence. Using our framework, we provide likely mechanistic insights into the physical control of loop dynamics. Specifically, we find that the strength of the loop extrusion anchorage site is largely governed by the amount and binding affinity of CTCF sites at the boundaries. Furthermore, we reveal that loop stability is regulated by the amount of intra-loop CTCF binding sites, where fewer sites within the loop lead to a more stable domain. Using targeted, single-nucleotide edit simulations with Chiron3D, we show that both loop strength and stability can be precisely controlled. Together, these results provide novel mechanistic insights into the physical control of genome organization and highlight the potential of decoding the DNA sequence logic in silico. AvailabilityThe Chiron3D pipeline is made available at https://github.com/BoevaLab/Chiron3D. Supp. informationSupp. data are available at Journal Name online.

genomics↗

bioRxiv2026

Evolutionary conservation and innovations of RNA polymerase II transcription elongation factors

In eukaryotes, transcription elongation factors (TEFs) associate with RNA Polymerase II (RNAPII) to facilitate gene expression and couple transcription to co-transcriptional processes, including chromatin regulation and RNA processing. To further our understanding of TEF biology, we developed a domain-centric analysis pipeline to perform a broad survey of ten TEF orthologs -- Paf1, Ctr9, Cdc73, Rtf1, Leo1, Spt4, Spt5, Spt6, Spn1, and Elf1 -- across the Tree of Life and analyze their evolutionary patterns in a structural context. We report evidence for all ten TEFs being present in the last eukaryotic common ancestor, indicating that mechanisms of TEF-mediated transcription regulation are both ancient and conserved. However, some early-diverging eukaryotic clades exhibit signs of altered TEF domain composition. A comparative phylogenetic analysis highlighted conserved regions of TEFs that are detected in both metazoans and fungi and other regions that appear clade-specific, detected only in metazoans. These observations, together with additional insights generated from evolutionary rate covariation analysis, shed light on under-characterized aspects of TEFs, including domains for which functions have yet to be dissected.

evolutionary biology↗

bioRxiv2026

TMEM33 deletion potentiates anti-tumor CD8+ T cell immunity

Improving responses to cancer immunotherapies requires deeper insight into the cellular mechanisms governing T cell-mediated anti-tumor immunity. TMEM33 is an endoplasmic reticulum-resident transmembrane protein enriched across multiple tumor types, with reported functions in anti-viral immunity as well as calcium and lipid homeostasis, yet its role in tumor immunosurveillance remains unknown. Using murine genetic models, we demonstrate that host TMEM33 constrains anti-tumor CD8+ T cell responses. Constitutive Tmem33-/- mice exhibited delayed melanoma tumor growth and increased CD8+ T cell infiltration. Antigen-specific CD8+ compartments in tumors of Tmem33-/- mice showed TCF-1+PD-1+ progenitor-exhausted cell (Tpex) enrichment, elevated effector function and reduced exhaustion, alongside improved effector memory expansion and T-bet expression in draining lymph nodes. We highlight that TMEM33 functions intrinsically within the T cell compartment, as TMEM33 deletion (1) enhanced polyclonal activation of naive CD8+ T cells ex vivo, (2) promoted preferential Tpex accumulation among adoptively transferred naive OT-I cells in B16F10-OVA tumors and draining lymph nodes, and (3) improved the potency of ex vivo-expanded OT-I cells in controlling tumor growth during adoptive cell therapy. Finally, in a large, prospectively recruited metastatic melanoma cohort, lower TMEM33 expression in patient CD8+ T cells significantly correlated with improved survival and elevated TCF-7 (encoding TCF-1). Collectively, our findings define TMEM33 as a formerly unrecognized intrinsic determinant of tumor-directed CD8+ T cell fate that limits Tpex maintenance, and restrains cell therapy responses, suggesting that its modulation may strengthen immunotherapeutic efficacy. One sentence summaryTMEM33 intrinsically limits progenitor exhausted CD8+ T cells, scales anti-tumor responses and predicts melanoma patient survival.

immunology↗

bioRxiv2025

REV1 inhibition enhances trinucleotide repeat mutagenesis

Trinucleotide repeat (TNR) instability has been implicated in the pathogenesis of numerous neurodegenerative disorders. Because TNR instability causes mutagenesis of the underlying gene, we refer to the repeat instability phenomenon as TNR mutagenesis in this study. While germline expansions destabilize TNR to cause disease anticipation, somatic cell TNR instability drives earlier onset of symptoms and further disease progression. However, the drivers behind these repeat length changes remain unclear. Current models suggest that DNA replication slippage events and the action of genome instability pathways, such as DNA repair, cause TNR mutagenesis. Whether mutagenic polymerases from the translesion synthesis (TLS) pathway result in TNR instability is unclear. TLS polymerases are best at bypassing difficult-to-replicate DNA regions due to bulky lesions or gaps in DNA. While some effects of TLS polymerases on TNR instability have been explored in lower organisms, evidence in human cells is lacking. Using a quantitative GFP reporter with expanded CAG repeats, we show that inhibition of the TLS polymerase REV1 by its inhibitor, JH-RE-06, or siRNA knockdown increases TNR instability and the underlying mutability. These results suggest that REV1 protects Trinucleotide repeat length mutagenesis through potential continuous DNA synthesis when replicative polymerases stall ahead of repeat secondary structures. Collectively, we present evidence of the role of the TLS pathway in TNR instability, with potential implications for understanding mutability mechanisms, disease biology, and therapeutic targeting.

genetics↗

bioRxiv2025

Evaluation of deep learning approaches for high-resolution chromatin accessibility prediction from genomic sequence

Fine-grained prediction of chromatin accessibility from DNA sequence is a foundational step in modeling gene expression changes resulting from sequence variants. Yet, few methods operate at the resolution necessary to capture subtle effects of single-nucleotide changes. Furthermore, it remains unclear which architectural components--such as residual connections, normalization strategies, or attention mechanisms--drive performance in these high-resolution predictions. To address these knowledge gaps, we systematically evaluate classic architectural choices and introduce ConvNeXt V2 blocks, originally developed for computer vision, as high-resolution feature extractors in deep learning models for genomic data. Integrated into diverse architectures--CNNs, LSTMs, dilated CNNs, and transformers--ConvNeXt V2 blocks consistently improve performance, leading to similar prediction accuracy across these different model types. This reveals that early feature extraction, rather than downstream architecture, is the primary determinant of prediction accuracy. A comprehensive evaluation of these models on ATAC-seq signal prediction at 4 bp resolution in a cell type-specific manner identifies the ConvNeXtbased dilated CNN as the most robust performer, better preserving the signals shape. Our codebase and benchmarks provide practical tools for high-resolution chromatin modeling.

genomics↗

bioRxiv2025

ERC 2.0 - evolutionary rate covariation update provides more powerful inference of functional interactions across large phylogenies

Evolutionary Rate Covariation (ERC) is an established comparative genomics method that identifies sets of genes sharing patterns of sequence evolution, which suggests shared function. Whereas many functional predictions of ERC have been empirically validated, its predictive power has hitherto been limited by its inability to tackle the large numbers of species in contemporary comparative genomics datasets. This study introduces ERC2.0, an enhanced methodology for studying ERC across phylogenies with hundreds of species and tens of thousands of genes. ERC2.0 improves upon previous iterations of ERC in algorithm speed, normalizing for heteroskedasticity, and normalizing correlations via Fisher transformations. These improvements have resulted in greater statistical power to predict biological function. In exemplar yeast and mammalian datasets, we demonstrate that the predictive power of ERC2.0 is improved relative to the previous method, ERC1.0, and that further improvements are obtained by using larger yeast and mammalian phylogenies. We attribute the improvements to both the larger datasets and improved rate normalization. We demonstrate that ERC2.0 has high predictive accuracy for known annotations and can predict the functions of genes in non-model systems. Our findings underscore the potential for ERC2.0 to be used as a single-pass computational tool in candidate gene screening and functional predictions.

evolutionary biology↗

bioRxiv2024

UniversalEPI: Harnessing Attention Mechanisms to Decode Chromatin Interactions in Rare and Unexplored Cell Types

Enhancer-promoter interactions (EPIs) play a central role in gene regulation, but experimental techniques such as Hi-C for mapping these interactions remain costly and labor-intensive. Computational methods have been developed to predict EPIs in silico from DNA sequence and chromatin information; however, there are major challenges with the generalizability and accuracy of predictions by existing methods across cell types and conditions unseen during model training. We developed and validated UniversalEPI, an attention-based deep ensemble model that predicts EPIs up to 2 Mb apart using only DNA sequence and chromatin accessibility (ATAC-seq) data. Unlike models that reconstruct full Hi-C contact maps, UniversalEPI focuses on biologically relevant, sparse chromatin interactions between accessible regulatory elements. It generalizes across both bulk and single-cell ATAC-seq-derived pseudo-bulk datasets, delivering state-of-the-art performance while using fewer input modalities than existing approaches. By modeling predictive uncertainty, UniversalEPI enables statistically robust differential analysis of chromatin interactions across conditions. We demonstrate its utility by tracking dynamic EPIs during human macrophage activation and identifying regulatory differences between cancer cell states in esophageal adenocarcinoma. By providing precalculated Hi-C predictions for 157 ENCODE datasets, UniversalEPI expands the scope and applicability of in silico 3D genome modeling for studying gene regulation in development and disease.

genomics↗

bioRxiv2022

SurvBoard: Standardised Benchmarking for Multi-omics Cancer Survival Models

Multi-omics data, which include genomic, transcriptomic, epigenetic, and proteomic data, are gaining increasing importance for determining the clinical outcomes of cancer patients. Several recent studies have evaluated various multi-modal integration strategies for cancer survival prediction, highlighting the need for standardizing model performance results. Addressing this issue, we introduce SurvBoard, a benchmark framework that standardizes key experimental design choices. SurvBoard enables comparisons between single-cancer and pan-cancer data models and assesses the benefits of using patient data with missing modalities. We also address common pitfalls in preprocessing and validating multi-omics cancer survival models. We apply SurvBoard to several exemplary use cases, further confirming that statistical models tend to outperform deep learning methods, especially for metrics measuring survival function calibration. Moreover, most models exhibit better performance when trained in a pan-cancer context and can benefit from leveraging samples for which data of some omics modalities are missing. We provide a web service for model evaluation and to make our benchmark results easily accessible and viewable: https://www.survboard.science/. All code is available on GitHub: https://github.com/BoevaLab/survboard/. All benchmark outputs are available on Zenodo: https://zenodo.org/records/11066227. O_TEXTBOXKey MessagesO_LIWe introduce SurvBoard, a comprehensive benchmarking framework for the standardized evaluation of multi-omics cancer survival models. SurvBoard provides an easily accessible platform for the reproducible comparison of models trained on single-cancer and pan-cancer datasets. The platform addresses issues such as the impact of missing modalities and variability in experimental setups. SurvBoard integrates data from four major cancer programs-TCGA, ICGC, TARGET, and METABRIC-to ensure a comprehensive evaluation across diverse types of cancer and research centers. C_LIO_LISurvBoard results confirm that statistical models generally outperform deep learning models in survival function calibration. We also find that pan-cancer training enhances model performance and that models benefit from incorporating data with missing modalities. C_LIO_LISurvBoard includes a web service that allows researchers to submit models for benchmarking and evaluation. A leaderboard is accessible via https://survboard.science/ to promote transparency and the continuous assessment of models performance. C_LI C_TEXTBOX

bioinformatics↗

bioRxiv2022

Small gene networks can delineate immune cell states and characterize immunotherapy response in melanoma

BackgroundSingle-cell sequencing studies have elucidated some of the underlying mechanisms responsible for immune checkpoint inhibitor (ICI) response, but are difficult to implement as a general strategy or in a clinical diagnostic setting. In contrast, bulk RNAseq is now routine for both research and clinical applications. Therefore, our analysis extracts small transcription factor-directed co-expression networks (regulons) from single-cell RNA-seq data and uses them to deconvolute immune functional states from bulk RNA-seq data to characterize patient responses. MethodsRegulons were inferred in pre-treatment CD45+ cells from metastatic melanoma samples (n=19) treated with first-line ICI therapy (discovery dataset). A logistic regression-based classifier identified immune cell states associated with response, which were characterized according to differentially active, cell-state specific regulons. The complexity of these regulons was reduced and scored in bulk RNAseq melanoma samples from four independent studies (n=209, validation dataset). Patients were clustered according to their regulon scores, and the associations between cluster assignment, response, and survival were determined. Intercellular communication analysis of cell states was performed, and the resulting effector genes were analyzed by trajectory inference. ResultsRegulons preserved the information of gene expression data and accurately delineated immune cell phenotypes, despite reducing dimensionality by > 100-fold. Four cell states, termed exhausted T cells, monocyte lineage cells, memory T cells, and B cells, were associated with therapeutic responses in the discovery dataset. The cell states were characterized by seven differentially active and specific regulons that showed low specificity in non-immune cells. Four clusters with significantly different response outcomes (P <0.001) were identified in the bulk RNAseq validation cohort. An intercellular link between exhausted T cells and monocyte lineage cells was established, whereby their cell numbers were correlated, and exhausted T cells predicted prognosis as a function of monocyte lineage cell number. Analysis of ligand - receptor expression suggested that monocyte lineage cells drive exhausted T cells into terminal exhaustion through programs that regulate antigen presentation, chronic inflammation, and negative co-stimulation. ConclusionsRegulon-based characterization of cell states provides robust and functionally informative markers that can deconvolve bulk RNA-seq data to identify ICI responders.

immunology↗

bioRxiv2022

ABA-induced SA accumulation causes higher susceptibility in B. juncea as compared to tolerant genotypes against A. brassicae

Alternaria brassicae, a necrotrophic pathogen causes Alternaria blight in members of the Brassicaceae family. An extensive yield-oriented breeding strategy has rendered Indian mustard (Brassica juncea) susceptible to many fungal pathogens however, Alternaria blight is one of the most pressing challenges of all because it causes almost 40-60% yield loss. Variable degree of tolerance is reported in wild relatives of B. juncea, Sinapis alba (White mustard) and camelina sativa (False flax) have been reported to exhibit moderate and high tolerance respectively against A. brassicae. Phytohormones are the essential regulator of the intricate mechanism of plant defence response. The hormones salicylic acid (SA) and jasmonic acid (JA) have been widely studied and recognized as important regulators of plant immune response. In the last decade, research has pointed out that other hormones like abscisic acid (ABA) also participate equally in plant defence. However, the role of ABA in defence responses and its cross-talk with SA and JA has not been fully understood in terms of Brassica-A. brassicae system. In this investigation, three genotypes-B. juncea, S. alba and C. sativa were selected and their response to exogenous application of SA, JA and ABA and their combination with A. brassicae were studied. Disease assessment, gene expression analysis and quantitative estimation of phytohormones showed that the B. juncea exhibited a weak JA-mediated defence response against A. brassicae and synergy between SA-ABA shifted the signalling mechanism to SA-mediated response leading to susceptibility in B. juncea. Tolerant genotypes, S. alba and C. sativa exhibited a robust JA-mediated response against A. brassicae and JA-ABA was found antagonistic in Brassica-A. brassicae phyto-pathosystem.

plant biology↗

bioRxiv2022

ProtFinder: finding subcellular locations of proteins using protein interaction networks

AO_SCPLOWBSTRACTC_SCPLOWProtein subcellular localization prediction plays a crucial role in improving our understanding of different diseases and consequently assists in building drug targeting and drug development pipelines. Proteins are known to co-exist at multiple subcellular locations which make the task of prediction extremely challenging. A protein interaction network is a graph that captures interactions between different proteins. It is safe to assume that if two proteins are interacting, they must share some subcellular locations. With this regard, we propose ProtFinder - the first deep learning-based model that exclusively relies on protein interaction networks to predict the multiple subcellular locations of proteins. We also integrate biological priors like the cellular component of Gene Ontology to make ProtFinder a more biology-aware intelligent system. ProtFinder is trained and tested using the STRING and BioPlex databases whereas the annotations of proteins are obtained from the Human Protein Atlas. Our model obtained an AUC-ROC score of 90.00% and an MCC score of 83.42% on a held-out set of proteins. We also apply ProtFinder to annotate proteins that currently do not have confident location annotations. We observe that ProtFinder is able to confirm some of these unreliable location annotations, while in some cases complementing the existing databases with novel location annotations. The source code for ProtFinder is available at https://github.com/UCLouvain-CBIO/ProtFinder.

bioinformatics↗

bioRxiv2020

An insight into transcriptome of LlaNAC Gene Over- expressing Tobacco Plants

Here, we report a whole transcriptome analysis of LlaNAC gene (from Lepidium latifolium) containing transgenic tobacco line (NC10) and wild type (WT), to attain deeper knowledge into the downstream genes activated by the over-expressing transgene. Transcriptome sequencing of NC10 and WT samples generated huge data using Illumina platform. The maximum number of unigenes GO annotated were of Biological process (8988, 3209) followed by molecular function (5155, 2577) and cellular components (3826, 1583) for WT and NC10 samples respectively. KEGG Pathway analysis revealed the unigenes were enriched in different functional pathway categories. The unigenes whose products involved in carbohydrate metabolism, glycan metabolism, and secondary metabolites synthesis were more for NC10 library in comparison to WT. Greater variety of transcription factors were involved in transgenic than wild-type plants. Genes like, Copia-like retrotransposable element, Peroxidase 64-like, Peptidyl-prolyl cis-trans isomerise, Cytochrome P450, Lipoyl synthase, CBL-interacting serine/threonine-protein kinase 5-like etc. were found differentially expressed in both the samples. Promoter analysis of these differentially expressed genes have elements for defence and stress response, abscisic acid response, shoot specific expression and light response, etc. In summary this study reports the involvement of the overexpressed genes in the dual action of cold tolerance and biomass accumulation, as sugars participate in both of these activities of the cell.

molecular biology↗

bioRxiv2019

Resolving cellular systems by ultra-sensitive and economical single-cell transcriptome filtering

Single-cell transcriptomics has sensitivity limits that restrict low abundance transcript identification, affects clustering and introduce artefact. Here, we describe Constellation DropSeq (C-DropSeq), a molecular transcriptome filter that delivers two orders of magnitude sensitivity gains by maximising read utility while reducing sequencing depth and costs. The simple and powerful method is broadly compatible with library preparation routines and was demonstrated by identifying and characterizing the activation of rare dendritic cell sub-populations.

molecular biology↗