Ricardo Wurmus

This study uses large language models to extract genomic biomarkers from oncology trials, showing open-source models outperform closed-source ones and benefit from fine-tuning.

Flexynesis provides a versatile deep learning framework for integrating bulk multi-omics data, offering usability, adaptability, and benchmarking across diverse datasets.

The paper investigates whether HOT regions are genuine transcription factor binding sites or ChIP-seq artifacts, finding they likely result from technical biases and specific sequence features.

The paper identifies functional miRNA::mRNA interactions in Drosophila using AGO CLIP experiments and quantifies their impact on gene expression and translation.

The paper compares DNase-seq and ATAC-seq for open chromatin and TF footprinting, showing protocol-specific biases affect performance and reproducibility.

The paper presents reproducible genomics analysis pipelines using GNU Guix for RNA-seq, ChIP-seq, Bisulfite-seq, and single-cell RNA-seq, ensuring consistent results across different environments.