IIIT Hyderabad Publications |
|||||||||
|
Deep Reinforcement Learning for Molecular Inverse Problem of 2 Nuclear Magnetic Resonance Spectra to Molecular StructureAuthors: bhuvanesh sridharan,Sarvesh Mehta,Yashaswi Pathak,U Deva Priyakumar Journal: Journal of Physical Chemistry Letters, 2022 Pages: 1-10 Date: 2022-06-09 Report no: IIIT/TR/2022/19 AbstractSpectroscopy is the study of how matter interacts with electromagnetic 5 radiation. The spectra of any molecule are highly information-rich, yet the inverse relation of 6 spectra to the corresponding molecular structure is still an unsolved problem. Nuclear 7 magnetic resonance (NMR) spectroscopy is one such critical technique in the scientists’ 8 toolkit to characterize molecules. In this work, a novel machine learning framework is 9 proposed that attempts to solve this inverse problem by navigating the chemical space to find 10 the correct structure given an NMR spectra. The proposed framework uses a combination of 11 online Monte Carlo tree search (MCTS) and a set of graph convolution networks to build a 12 molecule iteratively. Our method can predict the structure of the molecule ∼80% of the time 13 in its top 3 guesses for molecules with <10 heavy atoms. We believe that the proposed 14 framework is a significant step in solving the inverse design problem of NMR spectra Full article: pdf Centre for Computational Natural Sciences and Bioinformatics |
||||||||
Copyright © 2009 - IIIT Hyderabad. All Rights Reserved. |