Deep Learning Protein Conformational Space with Convolutions and Latent Interpolations

Determining the different conformational states of a protein and the transition paths between them is key to fully understanding the relationship between biomolecular structure and function. This can be accomplished by sampling protein conformational space with molecular simulation methodologies. De...

Descripción completa

Guardado en:

Detalles Bibliográficos
Autores principales:	Venkata K. Ramaswamy, Samuel C. Musson, Chris G. Willcocks, Matteo T. Degiacomi
Formato:	article
Lenguaje:	EN
Publicado:	American Physical Society 2021
Materias:	Physics QC1-999
Acceso en línea:	https://doaj.org/article/82f7acafe19041bc9e2a053f42497bd8
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

id	oai:doaj.org-article:82f7acafe19041bc9e2a053f42497bd8
record_format	dspace
spelling	oai:doaj.org-article:82f7acafe19041bc9e2a053f42497bd82021-12-02T11:37:43ZDeep Learning Protein Conformational Space with Convolutions and Latent Interpolations10.1103/PhysRevX.11.0110522160-3308https://doaj.org/article/82f7acafe19041bc9e2a053f42497bd82021-03-01T00:00:00Zhttp://doi.org/10.1103/PhysRevX.11.011052http://doi.org/10.1103/PhysRevX.11.011052https://doaj.org/toc/2160-3308Determining the different conformational states of a protein and the transition paths between them is key to fully understanding the relationship between biomolecular structure and function. This can be accomplished by sampling protein conformational space with molecular simulation methodologies. Despite advances in computing hardware and sampling techniques, simulations always yield a discretized representation of this space, with transition states undersampled proportionally to their associated energy barrier. We present a convolutional neural network that learns a continuous conformational space representation from example structures, and loss functions that ensure intermediates between examples are physically plausible. We show that this network, trained with simulations of distinct protein states, can correctly predict a biologically relevant transition path, without any example on the path provided. We also show we can transfer features learned from one protein to others, which results in superior performances, and requires a surprisingly small number of training examples.Venkata K. RamaswamySamuel C. MussonChris G. WillcocksMatteo T. DegiacomiAmerican Physical SocietyarticlePhysicsQC1-999ENPhysical Review X, Vol 11, Iss 1, p 011052 (2021)
institution	DOAJ
collection	DOAJ
language	EN
topic	Physics QC1-999
spellingShingle	Physics QC1-999 Venkata K. Ramaswamy Samuel C. Musson Chris G. Willcocks Matteo T. Degiacomi Deep Learning Protein Conformational Space with Convolutions and Latent Interpolations
description	Determining the different conformational states of a protein and the transition paths between them is key to fully understanding the relationship between biomolecular structure and function. This can be accomplished by sampling protein conformational space with molecular simulation methodologies. Despite advances in computing hardware and sampling techniques, simulations always yield a discretized representation of this space, with transition states undersampled proportionally to their associated energy barrier. We present a convolutional neural network that learns a continuous conformational space representation from example structures, and loss functions that ensure intermediates between examples are physically plausible. We show that this network, trained with simulations of distinct protein states, can correctly predict a biologically relevant transition path, without any example on the path provided. We also show we can transfer features learned from one protein to others, which results in superior performances, and requires a surprisingly small number of training examples.
format	article
author	Venkata K. Ramaswamy Samuel C. Musson Chris G. Willcocks Matteo T. Degiacomi
author_facet	Venkata K. Ramaswamy Samuel C. Musson Chris G. Willcocks Matteo T. Degiacomi
author_sort	Venkata K. Ramaswamy
title	Deep Learning Protein Conformational Space with Convolutions and Latent Interpolations
title_short	Deep Learning Protein Conformational Space with Convolutions and Latent Interpolations
title_full	Deep Learning Protein Conformational Space with Convolutions and Latent Interpolations
title_fullStr	Deep Learning Protein Conformational Space with Convolutions and Latent Interpolations
title_full_unstemmed	Deep Learning Protein Conformational Space with Convolutions and Latent Interpolations
title_sort	deep learning protein conformational space with convolutions and latent interpolations
publisher	American Physical Society
publishDate	2021
url	https://doaj.org/article/82f7acafe19041bc9e2a053f42497bd8
work_keys_str_mv	AT venkatakramaswamy deeplearningproteinconformationalspacewithconvolutionsandlatentinterpolations AT samuelcmusson deeplearningproteinconformationalspacewithconvolutionsandlatentinterpolations AT chrisgwillcocks deeplearningproteinconformationalspacewithconvolutionsandlatentinterpolations AT matteotdegiacomi deeplearningproteinconformationalspacewithconvolutionsandlatentinterpolations
_version_	1718395770483245056

Deep Learning Protein Conformational Space with Convolutions and Latent Interpolations

Ejemplares similares