Vladyslav Husak

FoldARE improves RNA secondary structure prediction by using pseudo-SHAPE constraints from structural ensembles, achieving high accuracy across benchmark datasets.

The paper evaluates ADMET prediction models on TDC leaderboards, finding most top models lack reproducibility and exhibit data leakage, with only three methods passing rigorous checks.

ArtiDock is a machine learning-based protein-ligand docking method that improves accuracy over classical techniques while maintaining low computational cost for high-throughput virtual screening.

This study evaluates ML-based conformational sampling in rigid protein-protein docking and finds it does not improve docking accuracy compared to using Apo structures.

This paper combines geometric pocket detection with large language models to prioritize protein binding pockets for drug discovery, using a curated dataset and automated pipeline.

This paper improves ADMET prediction by augmenting Mol2Vec embeddings with classical descriptors and feature selection, achieving top results on 10 of 16 TDC benchmarks.