If you enjoyed this talk, consider joining the Molecular Modeling and Drug Discovery (M2D2) talks live: m2d2.io/talks/m2d2/about/
Also consider joining the M2D2 Slack: m2d2group.slack.com/join/shar...
Title: DiffDock: Diffusion Steps, Twists, and Turns for Molecular Docking
Abstract: Predicting the binding structure of a small molecule ligand to a protein -- a task known as molecular docking -- is critical to drug design. Recent deep learning methods that treat docking as a regression problem have decreased runtime compared to traditional search-based methods but have yet to offer substantial improvements in accuracy. We instead frame molecular docking as a generative modeling problem and develop DiffDock, a diffusion generative model over the non-Euclidean manifold of ligand poses. To do so, we map this manifold to the product space of the degrees of freedom (translational, rotational, and torsional) involved in docking and develop an efficient diffusion process on this space. Empirically, DiffDock obtains a 38% top-1 success rate (RMSD less than 2A) on PDBBind, significantly outperforming the previous state-of-the-art of traditional docking (23%) and deep learning (20%) methods. Moreover, DiffDock has fast inference times and provides confidence estimates with high selective accuracy.
Paper - arxiv.org/abs/2210.01776
Speakers:
Hannes Stärk - / hannesstaerk
Gabriele Corso - / gabricorso
Bowen Jing - / bowen-jing
Twitter Prudencio: / tossouprudencio
Twitter Therence: / therence_mtl
Twitter Jonny: / hsu_jonny
Twitter Valence Discovery: / valence_ai
~
Chapters:
00:00 - Intro
02:06 - What is Molecular Docking?
07:32 - Problem with Deep Learning Models for Docking
12:11 - Diffusion Generative Models for docking
17:10 - Product Space Diffusion
19:16 - Score Model
20:40 - Workflow Overview
22:03 - Q+A
27:24 - Results
32:54 - Reverse Diffusion Process and Confidence Score Quality
41:40 - Q+A