Vincent Meyer

Vincent Meyer

Auto-generated profile, based on publicly available information

Vincent Meyer

Université Paris-Saclay, CEA, Centre National de Recherche en Génomique Humaine (CNRGH), Évry-Courcouronnes, France

Research Focus & Metrics

Academic Career Length

6Years (2021-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
2 May 2022 in JNCI: Journal of the National Cancer Institute
Verified
Journal Article
Open access
PDF file
AGAurélie A G Gabriel
Aurélie A G Gabriel
JAJoshua R Atkins
Joshua R Atkins
RPRicardo C C Penha
Ricardo C C Penha
KSKarl Smith-Byrne
Karl Smith-Byrne
VGValerie Gaborieau
Valerie Gaborieau
JMJames D McKay
James D McKay
Germline genetic variation contributes to lung cancer (LC) susceptibility. Previous genome-wide association studies (GWAS) have implicated susceptibility loci involved in smoking behaviors and DNA repair genes, but further work is required to identify susceptibility variants. To identify LC susceptibility loci, a family history-based genome-wide association by proxy (GWAx) of LC (48 843 European proxy LC patients, 195 387 controls) was combined with a previous LC GWAS (29 266 patients, 56 450 controls) by meta-analysis. Colocalization was used to explore candidate genes and overlap with existing traits at discovered susceptibility loci. Polygenic risk scores (PRS) were tested within an independent validation cohort (1 666 LC patients vs 6 664 controls) using variants selected from the LC susceptibility loci and a novel selection approach using published GWAS summary statistics. Finally, the effects of the LC PRS on somatic mutational burden were explored in patients whose tumor resections have been profiled by exome (n = 685) and genome sequencing (n = 61). Statistical tests were 2-sided. The GWAx–GWAS meta-analysis identified 8 novel LC loci. Colocalization implicated DNA repair genes (CHEK1), metabolic genes (CYP1A1), and smoking propensity genes (CHRNA4 and CHRNB2). PRS analysis demonstrated that these variants, as well as subgenome-wide significant variants related to expression quantitative trait loci and/or smoking propensity, assisted in LC genetic risk prediction (odds ratio = 1.37, 95% confidence interval = 1.29 to 1.45; P < .001). Patients with higher genetic PRS loads of smoking-related variants tended to have higher mutation burdens in their lung tumors. This study has expanded the number of LC susceptibility loci and provided insights into the molecular mechanisms by which these susceptibility variants contribute to LC development.

8 September 2025 in Bioinformatics
Verified
Journal Article
Open access
MMMallek Mziou-Sallami
Mallek Mziou-Sallami
PRPierrick Roger
Pierrick Roger
AGArnaud Gloaguen
Arnaud Gloaguen
CDClaire Dandine-Roulland
Claire Dandine-Roulland
TNThierry Jiogho Ngaho
Thierry Jiogho Ngaho
VMVincent Meyer
Vincent Meyer
Motivation: Graph Neural Network (GNN) models have emerged in many fields and notably for biological networks constituted by genes or proteins and their interactions. The majority of enrichment study methods apply over-representation analysis and gene/protein set scores according to the existing overlap between pathways. Such methods neglect knowledges coming from the interactions between the gene/protein sets. Here, we introduce a novel GNN-based enrichment analysis method called GNNenrich. GNNenrich, through multiple levels of embedding that integrate protein sequence properties and interactions network, establishes functional relationship to support biological interpretation. Results: GNNenrich have been tested and compared to Over-Representation Analysis technique (g: Profiler) and graph-based method (EnrichNet). It demonstrates the capacity to reproduce results provided by others approaches and offers new perspectives for interpretation, returning relevant results supported by Protein–Protein Interactions (PPIs). Availability and implementation: Source code is available at https://gitlab.com/cnrgh/gnn-enrich/gnn-enrich-article-demo.

29 December 2025 in Computational and Structural Biotechnology Journal
Verified
Journal Article
Open access
KPKarl Paygambar
Karl Paygambar
RJRoude Jean-Marie
Roude Jean-Marie
MMMallek Mziou-Sallami
Mallek Mziou-Sallami
VMVincent Meyer
Vincent Meyer
Optimizing genomic-based molecular subtyping is key to promote personalized medicine. While neural networks face setbacks regarding tabular data modeling, deep learning has undergone groundbreaking advances across multiple domains, catalyzing further breakthroughs across AI applications. New neural network architectures exhibit enhanced performance, efficiency, and robustness, which could benefit the genomic use-case.

27 November 2025
Verified
Conference Paper
Open access
CMCristina Marianini
Cristina Marianini
KPKarl Paygambar
Karl Paygambar
VMVincent Meyer
Vincent Meyer
MMMallek Mziou
Mallek Mziou

22 April 2026 in bioRxiv
Verified
Preprint
Open access
PDF file
ALAurélie Leduc
Aurélie Leduc
ABAsmae Bachr
Asmae Bachr
FSFlorian Sandron
Florian Sandron
MDMarc Delépine
Marc Delépine
DDDamien Delafoy
Damien Delafoy
VMVincent Meyer
Vincent Meyer
Background: Short read sequencing technologies have dominated the field of human whole genome sequencing in the past years in terms of cost, throughput, and accuracy. However, thanks to recent technological evolution, long read approaches have become increasingly competitive and complementary to short reads. With the gap in the cost per genome closing slowly between both approaches, long reads might replace short read sequencing in future research and clinical applications. Still, comprehensive evaluation is necessary to conclude on the performance and general advantages of each technology. Results: In this study, we compared the latest chemistries of major suppliers of short and long read technologies: Illumina short reads, Illumina Complete Long Reads (ICLR), Pacific Biosciences HiFi reads (PacBio), and Oxford Nanopore Technologies long reads (ONT). Using the HG002 human reference sample and established bioinformatics guidelines, we assessed their variant calling performance against the latest available truth sets at different levels of coverage. For single nucleotide variant detection, all technologies were equivalent. Despite the latest improvements in chemistry, indel calling with ONT continues to lag in accuracy behind other technologies. In contrast, long reads delivered a clear advantage in structural variant detection, surpassing short reads in both accuracy and sensitivity. The hybrid ICLR approach achieved intermediate performance, narrowing the gap between short and long read sequencing. Furthermore, long reads enhanced haplotype-phasing resolution, enabling the phasing of over 80% of the genome. Conclusions: These findings highlight the specific strengths and limitations of recent sequencing technologies, aiding the decision-making in future research projects, technological platforms development, and clinical applications.

30 June 2026
Verified
Conference Paper
Open access
KPKarl Paygambar
Karl Paygambar
ABAdda Akram Bendoukha
Adda Akram Bendoukha
OSOana Stan
Oana Stan
MMMallek Mziou
Mallek Mziou
VMVincent Meyer
Vincent Meyer
Machine learning holds the potential to improve personalized treatment strategies. Trust remains a key development axis towards adoption, as models have displayed unequal performance across subgroups in related applications. In this study, we investigate ancestry-related fairness, and evaluate alternative training strategies for PAM50 breast cancer subtyping with gene expression data. Using TCGA’s BRCA dataset, fairness is assessed regarding White, Black, Hispanic and Asian patients. Transformers, 2D and 1D convolutionnal neural networks, and XGBoost are integrated in this study, containing three training configurations. First, in a centralized baseline setting, disparities across models disfavor Asian and Hispanic samples, and are sustained after equalizing samples. Then, models are trained incorporating ancestry, which only marginally alters model behaviors, although minor improvements are observed for XGBoost and NeXt-TDNN regarding precision. Finally, federated learning, secured via fully homomorphic encryption, is evaluated through two strategies. The first reflects a realistic configuration, where data allocation across centers follows the geographical distribution, while the second uses a randomized distribution. Both setups lead to slightly better global performance, with group-wise results tied to sample distributions.