Molecular Geometry Prediction using a Deep Generative Graph Neural Network

Elman Mansimov, Omar Mahmood, Seokho Kang, Kyunghyun Cho

Research output: Contribution to journalArticle

Abstract

A molecule’s geometry, also known as conformation, is one of a molecule’s most important properties, determining the reactions it participates in, the bonds it forms, and the interactions it has with other molecules. Conventional conformation generation methods minimize hand-designed molecular force field energy functions that are often not well correlated with the true energy function of a molecule observed in nature. They generate geometrically diverse sets of conformations, some of which are very similar to the lowest-energy conformations and others of which are very different. In this paper, we propose a conditional deep generative graph neural network that learns an energy function by directly learning to generate molecular conformations that are energetically favorable and more likely to be observed experimentally in data-driven manner. On three large-scale datasets containing small molecules, we show that our method generates a set of conformations that on average is far more likely to be close to the corresponding reference conformations than are those obtained from conventional force field methods. Our method maintains geometrical diversity by generating conformations that are not too similar to each other, and is also computationally faster. We also show that our method can be used to provide initial coordinates for conventional force field methods. On one of the evaluated datasets we show that this combination allows us to combine the best of both methods, yielding generated conformations that are on average close to reference conformations with some very similar to reference conformations.

Original languageEnglish (US)
Article number20381
JournalScientific reports
Volume9
Issue number1
DOIs
StatePublished - Dec 1 2019

Fingerprint

Molecular Conformation
Hand
Learning
Datasets

ASJC Scopus subject areas

  • General

Cite this

Molecular Geometry Prediction using a Deep Generative Graph Neural Network. / Mansimov, Elman; Mahmood, Omar; Kang, Seokho; Cho, Kyunghyun.

In: Scientific reports, Vol. 9, No. 1, 20381, 01.12.2019.

Research output: Contribution to journalArticle

Mansimov, Elman ; Mahmood, Omar ; Kang, Seokho ; Cho, Kyunghyun. / Molecular Geometry Prediction using a Deep Generative Graph Neural Network. In: Scientific reports. 2019 ; Vol. 9, No. 1.
@article{66f909ae656c44d4aa0beafc181c2dd9,
title = "Molecular Geometry Prediction using a Deep Generative Graph Neural Network",
abstract = "A molecule’s geometry, also known as conformation, is one of a molecule’s most important properties, determining the reactions it participates in, the bonds it forms, and the interactions it has with other molecules. Conventional conformation generation methods minimize hand-designed molecular force field energy functions that are often not well correlated with the true energy function of a molecule observed in nature. They generate geometrically diverse sets of conformations, some of which are very similar to the lowest-energy conformations and others of which are very different. In this paper, we propose a conditional deep generative graph neural network that learns an energy function by directly learning to generate molecular conformations that are energetically favorable and more likely to be observed experimentally in data-driven manner. On three large-scale datasets containing small molecules, we show that our method generates a set of conformations that on average is far more likely to be close to the corresponding reference conformations than are those obtained from conventional force field methods. Our method maintains geometrical diversity by generating conformations that are not too similar to each other, and is also computationally faster. We also show that our method can be used to provide initial coordinates for conventional force field methods. On one of the evaluated datasets we show that this combination allows us to combine the best of both methods, yielding generated conformations that are on average close to reference conformations with some very similar to reference conformations.",
author = "Elman Mansimov and Omar Mahmood and Seokho Kang and Kyunghyun Cho",
year = "2019",
month = "12",
day = "1",
doi = "10.1038/s41598-019-56773-5",
language = "English (US)",
volume = "9",
journal = "Scientific Reports",
issn = "2045-2322",
publisher = "Nature Publishing Group",
number = "1",

}

TY - JOUR

T1 - Molecular Geometry Prediction using a Deep Generative Graph Neural Network

AU - Mansimov, Elman

AU - Mahmood, Omar

AU - Kang, Seokho

AU - Cho, Kyunghyun

PY - 2019/12/1

Y1 - 2019/12/1

N2 - A molecule’s geometry, also known as conformation, is one of a molecule’s most important properties, determining the reactions it participates in, the bonds it forms, and the interactions it has with other molecules. Conventional conformation generation methods minimize hand-designed molecular force field energy functions that are often not well correlated with the true energy function of a molecule observed in nature. They generate geometrically diverse sets of conformations, some of which are very similar to the lowest-energy conformations and others of which are very different. In this paper, we propose a conditional deep generative graph neural network that learns an energy function by directly learning to generate molecular conformations that are energetically favorable and more likely to be observed experimentally in data-driven manner. On three large-scale datasets containing small molecules, we show that our method generates a set of conformations that on average is far more likely to be close to the corresponding reference conformations than are those obtained from conventional force field methods. Our method maintains geometrical diversity by generating conformations that are not too similar to each other, and is also computationally faster. We also show that our method can be used to provide initial coordinates for conventional force field methods. On one of the evaluated datasets we show that this combination allows us to combine the best of both methods, yielding generated conformations that are on average close to reference conformations with some very similar to reference conformations.

AB - A molecule’s geometry, also known as conformation, is one of a molecule’s most important properties, determining the reactions it participates in, the bonds it forms, and the interactions it has with other molecules. Conventional conformation generation methods minimize hand-designed molecular force field energy functions that are often not well correlated with the true energy function of a molecule observed in nature. They generate geometrically diverse sets of conformations, some of which are very similar to the lowest-energy conformations and others of which are very different. In this paper, we propose a conditional deep generative graph neural network that learns an energy function by directly learning to generate molecular conformations that are energetically favorable and more likely to be observed experimentally in data-driven manner. On three large-scale datasets containing small molecules, we show that our method generates a set of conformations that on average is far more likely to be close to the corresponding reference conformations than are those obtained from conventional force field methods. Our method maintains geometrical diversity by generating conformations that are not too similar to each other, and is also computationally faster. We also show that our method can be used to provide initial coordinates for conventional force field methods. On one of the evaluated datasets we show that this combination allows us to combine the best of both methods, yielding generated conformations that are on average close to reference conformations with some very similar to reference conformations.

UR - http://www.scopus.com/inward/record.url?scp=85077180053&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85077180053&partnerID=8YFLogxK

U2 - 10.1038/s41598-019-56773-5

DO - 10.1038/s41598-019-56773-5

M3 - Article

C2 - 31892716

AN - SCOPUS:85077180053

VL - 9

JO - Scientific Reports

JF - Scientific Reports

SN - 2045-2322

IS - 1

M1 - 20381

ER -