Raphaël Poujol

This paper develops a deep learning framework for genomic ancestry prediction that is generalizable across diverse populations and interpretable using attribution techniques.

The paper models HLA diversity and viral evolution to assess T cell escape risk across populations, finding that escape mutations gain fitness with additional mutations.

The paper identifies cell-type-specific enhancer-gene interactions in cardiac cells and maps disease-associated SNPs to enhancers in endothelial cells.

The paper develops a workflow to improve the detection of intra-host SARS-CoV-2 mutations by reducing sequencing artifacts and batch effects using large-scale NGS data.

This study examines how temperature affects Spike-ACE2 interactions in Omicron subvariants and links these interactions to viral transmission efficiency.

This study identifies co-evolution and co-regulation in CYP3A and CYP4F genes across human populations, linking genetic diversity to drug metabolism and health outcomes.

This paper introduces the shortest reactional distance (SRD) metric to improve the biological interpretation of metabolite genome-wide association study (mGWAS) results by linking genes to metabolites based on metabolic pathway knowledge.

This study uses phylogenetic analysis and genome-wide association studies to investigate SARS-CoV-2 transmission between humans and animals, identifying species-specific mutations.

The paper computes local feature attributions for t-SNE using gradient-based methods and validates them on synthetic and biological datasets.

This study shows that immunosuppressed patients with hematologic cancers develop viral populations with immune escape mutations, suggesting the need for combination therapies.

The paper identifies immune evasion mechanisms in SARS-CoV-2 by mapping epitopes and analyzing viral genome sequences to understand evolutionary pressures and immune responses.

The paper compares protein aggregation propensity and mutation tolerance between naked mole-rat and mouse orthologous proteins, finding significant differences in specific protein subsets related to aging processes.

The paper presents computational tools for analyzing SARS-CoV-2 genetic data, showing that haplotype networks and unsupervised learning can reveal lineage emergence and population dynamics.

ImputeCoVNet is a 2D ResNet Autoencoder for imputing missing SARS-CoV-2 variants, achieving accurate results at low allele frequencies with faster computation than Hamming distance methods.