About BAGIM

BAGIM is an active community of Boston area scientists bringing together people from diverse fields of modeling and informatics to impact life and health sciences. BAGIM strives to create a forum for great scientific discussions covering a wide range of topics including data management, visualization, computational chemistry, drug discovery, protein structure, molecular modeling, structure-based drug design, data mining, software tools, and the sharing of goals and experiences. Our community is made up of participants from academia, government, and industry whose goal is to engage in the discussion of science involving a synthesis of theory and technology. Discussions sponsored by BAGIM are targeted to the needs and interests of informatics scientists, computational chemists, medicinal chemists, and statisticians. BAGIM also provides opportunities for networking within these disciplines as well as an arena for the dissemination of information of specific interest to the membership.

Friday, December 2, 2022

EquiBind: Geometric Deep Learning for Drug Binding Structure Prediction

Thursday, April 14, 2022 at 12:00 PM

Speaker:

Octavian Ganea

Details

Geometric Deep Learning Models for Predicting 3D Structures and Interactions of Molecules

Abstract:

Understanding 3D structures and interactions of biological nano-machines, such as proteins or drug-like molecules, is crucial for assisting drug and therapeutics discovery. A core problem is molecular docking, i.e., determining how two proteins or a protein and a drug-molecule attach and create a molecular complex. Having access to very fast computational docking tools would enable applications such as fast virtual search for drugs inhibiting disease proteins, in silico molecular design, or rapid drug side-effect prediction. However, existing computer models follow a very time-consuming strategy of sampling a large number (e.g., millions) of molecular complex candidates, followed by scoring, ranking, and fine-tuning steps. In this talk, I will show that geometry and deep learning (DL) can significantly reduce the enormous search space associated with the docking and molecular conformation problems. I will present my recent DL architectures, EquiDock and EquiBind, that perform a direct shot prediction of the molecular complex, and GeoMol, that models molecular flexibility. I will argue that the governing laws of geometry, physics, or chemistry that naturally constrain these 3D structures should be incorporated in DL solutions in a mathematically meaningful way. I will explain our key modeling concepts such as SE(3)-equivariant graph matching networks, attention keypoint sets, optimal transport for binding pocket prediction, and torsion angle neural networks. These approaches reduce the inference runtimes of open-source or commercial software from tens of minutes or hours to a few seconds, while being competitive or better in terms of quality. Finally, I will highlight a number of exciting on-going and future efforts in the space of artificial intelligence for structural biology and chemistry.

Bio:

Octavian Ganea is a postdoctoral researcher at CSAIL-MIT working with Tommi Jaakkola and Regina Barzilay on deep learning solutions for drug discovery and structural biology using geometric and physical inductive biases. He is part of and contributes to the Machine Learning for Pharmaceutical Discovery and Synthesis consortium, the Abdul Latif Jameel Clinic for Machine Learning in Health, the DARPA Accelerated Molecular Discovery program, and the ELLIS society. Octavian received his PhD from ETH Zurich under the supervision of Thomas Hofmann working on non-Euclidean representation learning for graphs, hierarchical data, and natural language processing. His published research includes a spotlight at ICLR 2022, spotlights at NeurIPS 2021 and 2018, and oral talks at ICML 2018 and 2019.

No comments:

Post a Comment