Known Molecule
Research on known molecules focuses on developing efficient methods for generating, optimizing, and predicting their properties, primarily to accelerate drug discovery and materials science. Current efforts leverage machine learning, employing architectures like graph neural networks, transformers, and diffusion models, often incorporating 3D structural information and multi-fidelity approaches to improve accuracy and efficiency. These advancements enable more rapid exploration of chemical space, leading to improved predictions of molecular properties and the design of molecules with desired characteristics, ultimately impacting drug development and materials design.
Papers
Unsupervised Learning of Molecular Embeddings for Enhanced Clustering and Emergent Properties for Chemical Compounds
Jaiveer Gill, Ratul Chakraborty, Reetham Gubba, Amy Liu, Shrey Jain, Chirag Iyer, Obaid Khwaja, Saurav Kumar
From Molecules to Materials: Pre-training Large Generalizable Models for Atomic Property Prediction
Nima Shoghi, Adeesh Kolluru, John R. Kitchin, Zachary W. Ulissi, C. Lawrence Zitnick, Brandon M. Wood
Predicting Drug Solubility Using Different Machine Learning Methods -- Linear Regression Model with Extracted Chemical Features vs Graph Convolutional Neural Network
John Ho, Zhao-Heng Yin, Colin Zhang, Nicole Guo, Yang Ha
Recovering a Molecule's 3D Dynamics from Liquid-phase Electron Microscopy Movies
Enze Ye, Yuhang Wang, Hong Zhang, Yiqin Gao, Huan Wang, He Sun