Сергей Андреевич Вилов

This study evaluates foundation models on human 3'UTR sequences, finding that 3UTR-specific models outperform genome-wide models in three of four RNA biology tasks.

The paper develops a machine learning model to predict promoter proximal Pol II pausing using chromatin and transcriptome data, identifies novel regulators, and quantifies their effects on gene regulation.

This paper introduces a neural network method for somatic SNP calling in WGS tumor samples without a matched normal, showing effectiveness across 5 cancer types.