DISSECT: deep semi-supervised consistency regularization for accurate cell type fraction and gene expression estimation

Cell deconvolution is the estimation of cell type fractions and cell type-specific gene expression from mixed data. An unmet challenge in cell deconvolution is the scarcity of realistic training data and the domain shift often observed in synthetic training data. Here, we show that two novel deep ne...

Full description

Saved in:
Bibliographic Details
Published inGenome Biology Vol. 25; no. 1; pp. 112 - 23
Main Authors Khatri, Robin, Machart, Pierre, Bonn, Stefan
Format Journal Article
LanguageEnglish
Published England BioMed Central 30.04.2024
BMC
Subjects
Algorithms
Cell deconvolution
Datasets
Deep Learning
Gene expression
Gene Expression Profiling - methods
Humans
Hypotheses
Method
Neural networks
Neural Networks, Computer
Probability distribution
Proteomics
Proteomics - methods
Semi-supervised learning
Simulation
Deep learning
Cell deconvolution
Semi-supervised learning
Online AccessGet full text

Cover

More Information
Summary:Cell deconvolution is the estimation of cell type fractions and cell type-specific gene expression from mixed data. An unmet challenge in cell deconvolution is the scarcity of realistic training data and the domain shift often observed in synthetic training data. Here, we show that two novel deep neural networks with simultaneous consistency regularization of the target and training domains significantly improve deconvolution performance. Our algorithm, DISSECT, outperforms competing algorithms in cell fraction and gene expression estimation by up to 14 percentage points. DISSECT can be easily adapted to other biomedical data types, as exemplified by our proteomic deconvolution experiments.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 23
ISSN:1474-760X
1474-7596
1474-760X
DOI:10.1186/s13059-024-03251-5