DecOT: Bulk Deconvolution With Optimal Transport Loss Using a Single-Cell Reference.

Gan Liu,Liang Ma,Xiuqin Liu

doi:10.3389/fgene.2022.825896

Abstract

Tissues are constituted of heterogeneous cell types. Although single-cell RNA sequencing has paved the way to a deeper understanding of organismal cellular composition, the high cost and technical noise have prevented its wide application. As an alternative, computational deconvolution of bulk tissues can be a cost-effective solution. In this study, we propose DecOT, a deconvolution method that uses the Wasserstein distance as a loss and applies scRNA-seq data as references to characterize the cell type composition from bulk tissue RNA-seq data. The Wasserstein loss in DecOT is able to utilize additional information from gene space. DecOT also applies an ensemble framework to integrate deconvolution results from multiple individuals’ references to mitigate the individual/batch effect. By benchmarking DecOT with four recently proposed square loss-based methods on pseudo-bulk data from four different single-cell data sets and real pancreatic islet bulk samples, we show that DecOT outperforms other methods and the ensemble framework is robust to the choice of references.

Highlights

Quantification of gene expression changes in different tissues or under different conditions gives information on how genes are regulated in organisms
We propose DecOT, which applies Wasserstein loss to estimate the relative abundance of cell types within a bulk sample by using a scRNA-seq reference ensemble of multi-individuals
We proposed DecOT, which applies single-cell data as references and uses Wasserstein distance as a loss function for decomposing bulk cell types

Summary

Introduction

Quantification of gene expression changes in different tissues or under different conditions gives information on how genes are regulated in organisms. The analysis of gene expression by using RNA sequencing (RNA-seq) has contributed substantially, since its development more than a decade ago, to our understanding of biological processes such as organism development, human disease progression, and patients’ response to treatments. The classic RNA-seq applied to bulk tissue samples has accumulated a rich reservoir of data sets, for example, GTEx, TCGA, and so forth (Tomczak et al, 2015, Carithers et al, 2015). Since tissues are heterogeneous, which comprise a variety of cell types, the bulk sequencing data only measure the average state of the mixed cell populations. When developing diagnostic techniques, such information would enable researchers to track the contribution of each cellular component during disease progressions (Schelker et al, 2017)

Objectives

Methods

Results

Conclusion