Detection of molecular behavior that characterizes systems using a deep learning approach

Katsuhiro Endo, Daisuke Yuhara, Katsufumi Tomobe, Kenji Yasuoka

Research output: Contribution to journalArticle

Abstract

Molecular dynamics (MD) simulation is a powerful computational method to observe molecular behavior. Although the detection of molecular behavior that characterizes systems is an important task in the study of MD, it is typically difficult and depends on human expert knowledge. Therefore, we propose a novel analysis scheme for MD data using deep neural networks. A key aspect of our scheme is the estimation of statistical distances between different ensembles that are probability distributions over the possible states of systems. This allows us to build low-dimensional embeddings of ensembles to visualize differences between systems in a compact metric space. Furthermore, the molecular behavior that contributes to the differences between systems can also be detected using the trained function of deep neural networks. The applicability of our scheme is demonstrated using three types of MD data. Our scheme could be a powerful tool to clarify the underlying physics in the molecular systems.

Original languageEnglish
Pages (from-to)10064-10071
Number of pages8
JournalNanoscale
Volume11
Issue number20
DOIs
Publication statusPublished - 2019 May 28

Fingerprint

Molecular dynamics
Computational methods
Probability distributions
Physics
Deep learning
Computer simulation
Deep neural networks

ASJC Scopus subject areas

  • Materials Science(all)

Cite this

Detection of molecular behavior that characterizes systems using a deep learning approach. / Endo, Katsuhiro; Yuhara, Daisuke; Tomobe, Katsufumi; Yasuoka, Kenji.

In: Nanoscale, Vol. 11, No. 20, 28.05.2019, p. 10064-10071.

Research output: Contribution to journalArticle

Endo, Katsuhiro ; Yuhara, Daisuke ; Tomobe, Katsufumi ; Yasuoka, Kenji. / Detection of molecular behavior that characterizes systems using a deep learning approach. In: Nanoscale. 2019 ; Vol. 11, No. 20. pp. 10064-10071.
@article{698174cdacf541009ca180c33eb1c4f0,
title = "Detection of molecular behavior that characterizes systems using a deep learning approach",
abstract = "Molecular dynamics (MD) simulation is a powerful computational method to observe molecular behavior. Although the detection of molecular behavior that characterizes systems is an important task in the study of MD, it is typically difficult and depends on human expert knowledge. Therefore, we propose a novel analysis scheme for MD data using deep neural networks. A key aspect of our scheme is the estimation of statistical distances between different ensembles that are probability distributions over the possible states of systems. This allows us to build low-dimensional embeddings of ensembles to visualize differences between systems in a compact metric space. Furthermore, the molecular behavior that contributes to the differences between systems can also be detected using the trained function of deep neural networks. The applicability of our scheme is demonstrated using three types of MD data. Our scheme could be a powerful tool to clarify the underlying physics in the molecular systems.",
author = "Katsuhiro Endo and Daisuke Yuhara and Katsufumi Tomobe and Kenji Yasuoka",
year = "2019",
month = "5",
day = "28",
doi = "10.1039/c9nr00219g",
language = "English",
volume = "11",
pages = "10064--10071",
journal = "Nanoscale",
issn = "2040-3364",
publisher = "Royal Society of Chemistry",
number = "20",

}

TY - JOUR

T1 - Detection of molecular behavior that characterizes systems using a deep learning approach

AU - Endo, Katsuhiro

AU - Yuhara, Daisuke

AU - Tomobe, Katsufumi

AU - Yasuoka, Kenji

PY - 2019/5/28

Y1 - 2019/5/28

N2 - Molecular dynamics (MD) simulation is a powerful computational method to observe molecular behavior. Although the detection of molecular behavior that characterizes systems is an important task in the study of MD, it is typically difficult and depends on human expert knowledge. Therefore, we propose a novel analysis scheme for MD data using deep neural networks. A key aspect of our scheme is the estimation of statistical distances between different ensembles that are probability distributions over the possible states of systems. This allows us to build low-dimensional embeddings of ensembles to visualize differences between systems in a compact metric space. Furthermore, the molecular behavior that contributes to the differences between systems can also be detected using the trained function of deep neural networks. The applicability of our scheme is demonstrated using three types of MD data. Our scheme could be a powerful tool to clarify the underlying physics in the molecular systems.

AB - Molecular dynamics (MD) simulation is a powerful computational method to observe molecular behavior. Although the detection of molecular behavior that characterizes systems is an important task in the study of MD, it is typically difficult and depends on human expert knowledge. Therefore, we propose a novel analysis scheme for MD data using deep neural networks. A key aspect of our scheme is the estimation of statistical distances between different ensembles that are probability distributions over the possible states of systems. This allows us to build low-dimensional embeddings of ensembles to visualize differences between systems in a compact metric space. Furthermore, the molecular behavior that contributes to the differences between systems can also be detected using the trained function of deep neural networks. The applicability of our scheme is demonstrated using three types of MD data. Our scheme could be a powerful tool to clarify the underlying physics in the molecular systems.

UR - http://www.scopus.com/inward/record.url?scp=85066042504&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85066042504&partnerID=8YFLogxK

U2 - 10.1039/c9nr00219g

DO - 10.1039/c9nr00219g

M3 - Article

VL - 11

SP - 10064

EP - 10071

JO - Nanoscale

JF - Nanoscale

SN - 2040-3364

IS - 20

ER -