Timezone: »
Molecular shape and geometry dictate key biophysical recognition processes, yet many modern graph neural networks disregard 3D information for molecular property prediction. Here, we propose a new contrastive-learning procedure for graph neural networks, Molecular Contrastive Learning from Shape Similarity (MolCLaSS), that implicitly learns a three-dimensional representation. Rather than directly encoding or targeting three-dimensional poses, MolCLaSS matches a similarity objective based on Gaussian overlays to learn a meaningful representation of molecular shape. We demonstrate how this framework naturally captures key aspects of three-dimensionality that two-dimensional representations cannot and provides an inductive framework for scaffold hopping.
Author Information
Austin Atsango (Stanford University)
Nathaniel Diamant (genentech)
Ziqing Lu
Tommaso Biancalani (Genentech)
Gabriele Scalia (Genentech)
Kangway Chuang (Genentech Research and Early Development)
More from the Same Authors
-
2022 : STab: Self-supervised Learning for Tabular Data »
Ehsan Hajiramezanali · Max Shen · Gabriele Scalia · Nathaniel Diamant -
2022 : Biological Cartography: Building and Benchmarking Representations of Life »
Safiye Celik · Jan-Christian Huetter · Sandra Melo · Nathan Lazar · Rahul Mohan · Conor Tillinghast · Tommaso Biancalani · Marta Fay · Berton Earnshaw · Imran Haque -
2022 : Deep Fitness Inference for Drug Discovery with Directed Evolution »
Nathaniel Diamant · Ziqing Lu · Christina Helmling · Kangway Chuang · Christian Cunningham · Tommaso Biancalani · Gabriele Scalia · Max Shen -
2022 : Deep Fitness Inference for Drug Discovery with Directed Evolution »
Nathaniel Diamant · Ziqing Lu · Christina Helmling · Kangway Chuang · Christian Cunningham · Tommaso Biancalani · Gabriele Scalia · Max Shen