Reproduce “Matcha: Multi-Stage Riemannian Flow Matching for Accurate and Physically Valid Molecular Docking” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Daria Frolova, Talgat Daulbaev, Egor Sevriugov, Sergei A. Nikolenko, Dmitry N. Ivankov, Ivan Oseledets, Marina A. Pak

Matcha predicts ligand binding poses in protein pockets through three separately trained stages that locate, orient, and refine flexible-bond rotations, then uses GNINA to minimize, filter, and rank candidates. In blind docking on Astex, 89.2 ± 1.5% of poses met both symmetry-corrected heavy-atom RMSD ≤ 2 Å and PoseBusters validity criteria. These benchmark results concern holo receptors, not experimental binding.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.