Biology subjects

Vermeirssen, V.

Publications and source records attributed to Vermeirssen, V..

10 recordsLinked to original sources

bioRxiv2026

Focus on the edges: a biomolecular network of histone PTMs, metabolites and proteins unveils functional entanglement in AML

The cell phenotype is not a direct manifestation of the genotype but rather a product of cellular history and the environmental context. However, individual biomolecules cannot change independently and show coordinated behavior. To study this in acute myeloid leukemia (AML), we built a unique multi-omics biomolecular network made from proteins, metabolites and histone posttranslational modifications (hPTMs) sequentially extracted from each cell pellet. Edges between the nodes are measured directly using 400 LC-MSMS runs that cover 18 AML cell lines. We provide a novel conceptual framework to illustrate the different classes of functional entanglement between and within omics layers and present the data in three interactive data browsers to allow full community access. To help navigate the network, we approach it from the perspective of two biomolecular targets, i.e. CD34 and the epigenetic mark Histone H3 lysine 27 trimethylation (H3K27me3). Now, this easily accessible biomolecular network serves as a starting point for building and testing hypotheses and streamlining drug development, in the process positioning biomolecular associations center stage in understanding phenotypic complexity.

cancer biology↗

bioRxiv2026

Lemonite: identification of regulatory metabolites through data-driven, interpretable integration of transcriptomics and metabolomics data

Current transcriptomics-metabolomics integration approaches are either limited by poor interpretability or constrained by incomplete prior knowledge, preventing the systematic identification of regulatory metabolites. Here, we present Lemonite, a data-driven and interpretable framework for integrating bulk transcriptomics and metabolomics data to uncover regulatory metabolites acting on gene modules. Lemonite extends module network inference to jointly associate transcription factors and metabolites with gene programs, without requiring prior differential analysis or complete metabolome annotation. To contextualize predictions, we constructed a comprehensive gene/protein-metabolite knowledge graph integrating over 370 000 metabolite-gene/protein and 2.1 million protein-protein interactions. Applied to glioblastoma (n=99) and inflammatory bowel disease (n=75) cohorts, Lemonite identified over 50 functionally coherent gene modules per disease, revealing established and previously uncharacterized metabolite-gene regulatory relationships. In glioblastoma, myo-inositol and phosphatidylcholines, together with IRF6, regulate mesenchymal-like immune programs, which upon integration with single-cell transcriptomics are primarily expressed in tumor-associated macrophages and monocytes. In inflammatory bowel disease, regulatory metabolites were prioritized that change the expression of their predicted target genes in colonic epithelial cells in vitro. Overall, Lemonite provides a principled framework to explore the genome-wide regulatory potential of the metabolome and to generate biologically interpretable, experimentally testable hypotheses from multi-omics data.

bioinformatics↗

bioRxiv2026

Comparison of different computational frameworks for metabolic modeling from single-cell transcriptomics data in glioblastoma

Metabolic reprogramming is a hallmark of glioblastoma, yet how distinct malignant and tumor microenvironment cell populations contribute to this metabolic heterogeneity remains poorly defined. Since direct single-cell metabolomics remains technically limited, transcriptomics-based computational inference offers a powerful alternative. Here we apply and systematically compare three complementary computational methods: (1) metabolic pathway activity scoring, (2) gene regulatory network inference focused on metabolic enzyme gene regulation, and (3) single-cell metabolic flux prediction. These methods were applied to snRNA-seq data from a set of GBM patient samples using the Human1 genome-scale metabolic model as a unified reaction and pathway annotation prior knowledge reference. Across all three methods, tumor-associated macrophages emerge as the metabolically dominant tumor microenvironment population. Tumor-associated macrophages in mesenchymal-like tumors show coordinated transcriptional control of lipid metabolism by five recurrently active transcription factors. They also exhibit consistent nucleotide biosynthesis flux and glutamate-to-glutamine conversion potentially supporting malignant cells. These findings demonstrate that multi-layered metabolic inference can resolve cell-type/state-specific dependencies in glioblastoma and highlight tumor-associated macrophage metabolism as a promising therapeutic target

systems biology↗

bioRxiv2026

Evaluation of deep learning tools for chromatin contact prediction

BackgroundThree-dimensional chromatin organization plays a central role in gene regulation and is commonly measured using Hi-C technology. Recently, deep learning models have been developed to predict Hi-C contact maps from genomic and epigenomic features, offering a computational alternative to costly experimental assays. However, the performance, robustness, and biological interpretability of these models remain unclear due to the absence of systematic benchmarking. ResultsWe present a comprehensive benchmark to evaluate five Hi-C prediction models, C.Origami, Epiphany, ChromaFold, HiCDiffusion, and GRACHIP, across multiple evaluation criteria, including accuracy, visual fidelity and loop detection. Among all models, Epiphany achieved the strongest overall performance, combining high accuracy, cell-type generalization, realistic image quality and reliable loop detection. Moreover, we evaluated predicted contact maps using four different loop-callers to assess the impact of model choice on loop detection performance. Despite the coarse resolution, many models could recover biologically relevant interactions. Notably, structural map quality was more critical than the choice of loop-caller for reliable detection. Finally, ablation analyses revealed that epigenomic signals are influential features for accurate Hi-C prediction. Despite the use of multiple input modalities in many models, only a limited subset contributed substantially to predictive performance. ConclusionsThis study provides a systematic comparison of deep learning models for Hi-C prediction and highlights the importance of specific regulatory signals in reconstructing 3D chromatin organization. The proposed evaluation framework clarifies model behaviours and offers guidance for the development and interpretation of Hi-C prediction methods.

bioinformatics↗

bioRxiv2024

A human-specific, concerted repression of microcephaly genes contributes to radiation-induced growth defects in forebrain organoids

AbstractPrenatal radiation-induced DNA damage poses a significant threat to normal brain development, resulting in microcephaly which primarily affects the cerebral cortex. It is unclear which molecular mechanisms are at the basis of this defect in humans as the few mechanistic studies performed so far were done in animals. Here, we leveraged human embryonic stem cell- derived forebrain organoids as a model for human corticogenesis. Organoids were X-irradiated with a moderate and a high dose at different time points, representing very early and mid corticogenesis. Irradiation caused a dose- and developmental-timing-dependent reduction in organoid size, which was more prominent in developmentally younger organoids. This coincided with a dose-dependent canonical p53-DREAM-dependent DNA damage response (DDR), consisting of cell cycle arrest, DNA repair and apoptosis. The DDR was delayed and less pronounced in the older organoids. Besides the DDR, we observed radiation-induced premature differentiation of neural progenitors and changes in metabolism. Importantly, our transcriptomic analysis furthermore demonstrated a concerted p53-E2F4-dependent repression of primary microcephaly genes. We found that this was a human-specific feature, as it was not observed in mouse embryonic brains or primary mouse neural progenitor cells. Thus, human forebrain organoids are an excellent model to investigate prenatal DNA damage-induced microcephaly and to uncover potentially targetable human-specific pathways.

developmental biology↗

bioRxiv2023

Evaluation of single-sample network inference methods for precision oncology

A major challenge in precision oncology is to identify targetable cancer vulnerabilities in individual patients. Modelling high-throughput omics data in biological networks allows identifying key molecules and processes of tumorigenesis. Traditionally, network inference methods rely on many samples to contain sufficient information for learning and predicting gene interactions for a group of patients. However, to implement patient-tailored approaches in precision oncology, we need to interpret omics data at the level of the individual patient. Several single-sample network inference methods have been developed that infer biological networks for an individual sample from bulk RNA-seq data. However, only a limited comparison of these methods has been made. Moreover, many methods rely on normal tissue samples as reference point for the tumor samples, which is not always available. Here, we conducted an evaluation of the single-sample network inference methods SSN, LIONESS, iENA, CSN and SSPGI using expression profiles of lung and brain cancer cell lines from the CCLE database. The methods constructed networks with distinct network topologies, as observed by edge weight distributions and other network characteristics. Further, hub gene analyses revealed different degrees of subtype-specificity across methods. Single-sample networks were able to distinguish between tumor subtypes, as exemplified by edge weight clustering, enrichment of known subtype-specific driver genes among hub gene sets, and differential node importance. Finally, we show that single-sample networks correlate better to other omics data from the same cell line as compared to aggregate networks. Our results point to the important role of single-sample network inference in precision medicine.

systems biology↗

bioRxiv2023

Comparative gene regulatory network analysis in Alzheimer's disease and major depressive disorder

Alzheimers disease and major depressive disorder are prevalent, devastating conditions with limited treatment options. Recent insights suggest that despite distinct phenotypes, these disorders share similar processes such as neuroinflammation. However, the extent of overlapping biological processes and their underlying molecular mechanisms remain to be elucidated. Therefore, we adopted a computational systems biology approach to compare regulatory programs in the prefrontal cortex of both disorders. Leveraging publicly available RNA sequencing data on different human cohorts, both at bulk and single-cell level, and using diverse computational methodologies, we inferred gene regulatory networks, which model the molecular interactions between transcription factors and their target genes, and characterized cell-type-specific regulatory programs and biological pathways. We identified core regulatory circuits shared in both disorders, including transcription factors that play a pivotal role in microglial activation such as IKZF1, IRF8, RUNX1 and SPI1. Most of these transcription factors had a reported role in Alzheimers, but not in depression. We found several common pathways such as microglial activation, but also more disease-specific pathways. Through orthogonal data analysis, we were able to validate several of the predicted regulatory interactions in Alzheimers disease and major depressive disorder. In summary, our work revealed neuroinflammation and microglial activation in both diseases, under the control of shared core regulatory circuits. The potential relevance of these transcription factors and genes warrants additional investigation, especially in depression, offering possible novel therapeutic opportunities.

systems biology↗

bioRxiv2022

SUBATOMIC: a SUbgraph BAsed mulTi-OMIcs Clustering framework to analyze integrated multi-edge networks

Representing the complex interplay between different types of biomolecules across different omics layers in multi-omics networks bears great potential to gain a deep mechanistic understanding of gene regulation and disease. However, multi-omics networks easily grow into giant hairball structures that hamper biological interpretation. Module detection methods can decompose these networks into smaller interpretable modules. However, these methods are not adapted to deal with multi-omics data nor consider topological features. When deriving very large modules or ignoring the broader network context, interpretability remains limited. To address these issues, we developed a subgraph based multi-omics clustering framework (SUBATOMIC), which infers small and interpretable modules with a specific topology while keeping track of connections to other modules and regulators. SUBATOMIC groups specific molecular interactions in composite network subgraphs of two and three nodes and clusters them into topological modules. These are functionally annotated, visualized and overlaid with expression profiles to go from static to dynamic modules. To preserve the larger network context, SUBATOMIC investigates statistically the connections in between modules as well as between modules and regulators such as miRNAs and transcription factors. We applied SUBATOMIC to analyze a composite Homo sapiens network containing transcription factor-target gene, miRNA-target gene, protein-protein, homologous and co-functional interactions from different databases. We derived and annotated 5586 modules with diverse topological, functional and regulatory properties. We created novel functional hypotheses for unannotated genes. Furthermore, we integrated modules with condition specific expression data to study the influence of hypoxia in three cancer cell lines. We developed two prioritization strategies to identify the most relevant modules in specific biological contexts: one considering GO term enrichments and one calculating an activity score reflecting the degree of differential expression. Both strategies yielded modules specifically reacting to low oxygen levels. In conclusion, we developed the SUBATOMIC framework that generates interpretable modules from multi-omics networks and applied it to hypoxia in cancer. SUBATOMIC can infer and contextualize modules, explore condition or disease specific modules, identify regulators and functionally related modules, and derive novel gene functions for uncharacterized genes. The software is available at https://github.com/CBIGR/SUBATOMIC.

bioinformatics↗

bioRxiv2021

Automatic end-to-end analysis of high-throughput in vitro cell culture screening by HTSplotter

In biomedical research, high-throughput screening is often applied as it comes with automatization, higher-efficiency, and more and faster results. High-throughput screening experiments encompass drug, drug combination, genetic perturbagen or a combination of genetic and chemical perturbagen screens. These experiments are conducted in real-time assays over time or in an endpoint assay. The data analysis consists of data cleaning and structuring, as well as further data processing and visualisation, which, due to the amount of data, can easily become laborious, time consuming and error-prone. Therefore, several tools have been developed to aid researchers in this process, but these typically focus on specific experimental set-ups and are unable to process data of several time points and genetic-chemical perturbagen screens. To meet these needs, we developed HTSplotter, available as web tool and Python module, which performs automatic data analysis and visualisation of either endpoint or real-time assays from different high-throughput screening experiments: drug, drug combination, genetic perturbagen and genetic-chemical perturbagen screens. HTSplotter implements an algorithm based on conditional statements in order to identify experiment type and controls. After appropriate data normalization, including growth rate normalization, HTSplotter executes downstream analyses such as dose-response relationship and drug synergism assessment by the Bliss independence (BI), Zero Interaction Potency (ZIP) and Highest Single Agent (HAS) methods. All results are exported as a text file and plots are saved in a PDF file. The main advantage of HTSplotter over other available tools is the automatic analysis of genetic-chemical perturbagen screens and real-time assays where growth rate and perturbagen effect results are plotted over time. In conclusion, HTSplotter allows for the automatic end-to-end data processing, analysis and visualisation of various high-throughput in vitro cell culture screens, offering major improvements in terms of versatility, efficiency and time over existing tools.

bioinformatics↗

bioRxiv2020

RRM2 is a target for synthetic lethal interactions with replication stress checkpoint addiction in high-risk neuroblastoma

Neuroblastoma is a pediatric tumor originating from the sympathetic nervous system responsible for 10-15 percent of all childhood cancer deaths. Half of all neuroblastoma patients present with high-risk disease at diagnosis. Despite intensive multi-modal therapies nearly 50 percent of high-risk cases relapse and die of their disease. In contrast to the overall paucity of mutations, high-risk neuroblastoma nearly invariably present with recurrent somatic segmental chromosome copy number variants. For several focal aberrations (e.g. MYCN and LIN28B amplification), the direct role in tumor formation has been established. However, for recurrent aberrations, such as chromosome 2p and 17q gains, the identification of genes contributing to tumor initiation or progression has been challenging due to the scarcity of small segmental gains or amplifications. In this study, we identified and functionally evaluated the ribonucleotide reductase regulatory subunit 2 (RRM2) as a top-ranked 2p putative co-driver and therapeutic target in high-risk neuroblastoma enforcing replicative stress resistance. In vitro knock down and pharmacological RRM2 inhibition highlight RRM2 dependency in neuroblastoma cells, further supported by the finding that co-overexpression of RRM2 in a d{beta}h-MYCN transgenic zebrafish line increased tumor penetrance with 80% and accelerated tumor formation. Given the critical role of RRM2 in replication fork progression and regulation of RRM2 through ATR/CHK1 signaling, we tested combined RRM2 and ATR/CHK1 small molecule inhibition with triapine and BAY1895344/prexasertib respectively, and observed strong synergism, in particular for combined RRM2 and CHK1 inhibition. Transcriptome analysis following combinatorial drugging revealed HEXIM1 as one of the strongest upregulated genes. Using programmable DNA binding of dCas9 with a promiscuous biotin ligase, RRM2 promotor bound proteins were identified including HEXIM1 and NurRD complex members, supporting a cooperative role for HEXIM1 upregulation together with CHK1 inhibition in further attenuating RRM2 expression levels. We evaluated the impact of combined RRM2/CHK1 inhibition in vivo, with treatment of a murine xenograft model showing rapid and complete tumor regression, without tumor regrowth upon treatment arrest. In conclusion, we identified RRM2 as a novel dependency gene in neuroblastoma and promising target for synergistic drug combinations with small compounds targeting DNA checkpoint regulators.

cancer biology↗