VEGA is an interpretable generative model for inferring biological network activity in single-cell transcriptomics

VEGA is an interpretable generative model for inferring biological network activity in single-cell transcriptomics

Developing interpretable models is a major challenge in single cell deep learning. Here we show that the VEGA variational autoencoder model, whose decoder wiring mirrors gene modules, can provide direct interpretability to the latent space further enabling the inference of biological module activity...

Full description

Bibliographic Details
Main Authors: Lucas Seninge, Ioannis Anastopoulos, Hongxu Ding, Joshua Stuart
Format: Article
Language:English
Published: Nature Publishing Group 2021-09-01
Series:Nature Communications
Online Access:https://doi.org/10.1038/s41467-021-26017-0
id doaj-ed3aa6f7d98748388fc0f47ff767569f
record_format Article
spelling doaj-ed3aa6f7d98748388fc0f47ff767569f2021-10-03T11:49:41ZengNature Publishing GroupNature Communications2041-17232021-09-011211910.1038/s41467-021-26017-0VEGA is an interpretable generative model for inferring biological network activity in single-cell transcriptomicsLucas Seninge0Ioannis Anastopoulos1Hongxu Ding2Joshua Stuart3Department of Biomolecular Engineering and Genomics Institute, University of CaliforniaDepartment of Biomolecular Engineering and Genomics Institute, University of CaliforniaDepartment of Biomolecular Engineering and Genomics Institute, University of CaliforniaDepartment of Biomolecular Engineering and Genomics Institute, University of CaliforniaDeveloping interpretable models is a major challenge in single cell deep learning. Here we show that the VEGA variational autoencoder model, whose decoder wiring mirrors gene modules, can provide direct interpretability to the latent space further enabling the inference of biological module activity.https://doi.org/10.1038/s41467-021-26017-0
collection DOAJ
language English
format Article
sources DOAJ
author Lucas Seninge
Ioannis Anastopoulos
Hongxu Ding
Joshua Stuart
spellingShingle Lucas Seninge
Ioannis Anastopoulos
Hongxu Ding
Joshua Stuart
VEGA is an interpretable generative model for inferring biological network activity in single-cell transcriptomics
Nature Communications
author_facet Lucas Seninge
Ioannis Anastopoulos
Hongxu Ding
Joshua Stuart
author_sort Lucas Seninge
title VEGA is an interpretable generative model for inferring biological network activity in single-cell transcriptomics
title_short VEGA is an interpretable generative model for inferring biological network activity in single-cell transcriptomics
title_full VEGA is an interpretable generative model for inferring biological network activity in single-cell transcriptomics
title_fullStr VEGA is an interpretable generative model for inferring biological network activity in single-cell transcriptomics
title_full_unstemmed VEGA is an interpretable generative model for inferring biological network activity in single-cell transcriptomics
title_sort vega is an interpretable generative model for inferring biological network activity in single-cell transcriptomics
publisher Nature Publishing Group
series Nature Communications
issn 2041-1723
publishDate 2021-09-01
description Developing interpretable models is a major challenge in single cell deep learning. Here we show that the VEGA variational autoencoder model, whose decoder wiring mirrors gene modules, can provide direct interpretability to the latent space further enabling the inference of biological module activity.
url https://doi.org/10.1038/s41467-021-26017-0
work_keys_str_mv AT lucasseninge vegaisaninterpretablegenerativemodelforinferringbiologicalnetworkactivityinsinglecelltranscriptomics
AT ioannisanastopoulos vegaisaninterpretablegenerativemodelforinferringbiologicalnetworkactivityinsinglecelltranscriptomics
AT hongxuding vegaisaninterpretablegenerativemodelforinferringbiologicalnetworkactivityinsinglecelltranscriptomics
AT joshuastuart vegaisaninterpretablegenerativemodelforinferringbiologicalnetworkactivityinsinglecelltranscriptomics
_version_ 1716845236685111296

Similar Items