Interpretation of Heterogeneous Single-Cell Gene Expression Data

We develop a novel matrix factorization tool named 'scINSIGHT' to jointly analyze multiple single-cell gene expression samples from biologically heterogeneous sources, such as different disease phases, treatment groups, or developmental stages. Given multiple gene expression samples from different biological conditions, 'scINSIGHT' simultaneously identifies common and condition-specific gene modules and quantify their expression levels in each sample in a lower-dimensional space. With the factorized results, the inferred expression levels and memberships of common gene modules can be used to cluster cells and detect cell identities, and the condition-specific gene modules can help compare functional differences in transcriptomes from distinct conditions. Please also see Qian K, Fu SW, Li HW, Li WV (2022) .

scINSIGHT for interpreting single cell gene expression in biologically heterogeneous data

Kun Qian, Wei Vivian Li 2025-10-19

Latest News

2025/10/19:

• Version 0.1.5 released!

Introduction

scINSIGHT uses a novel matrix factorization model to jointly analyze multiple single-cell gene expression samples from biologically heterogeneous sources, such as different disease phases, treatment groups, or developmental stages. It assumes that each gene module is a sparse and non-negative linear combination of genes, and each cell is jointly defined by the expression of common and condition-specific modules. Given multiple gene expression samples from different biological conditions, scINSIGHT aims to simultaneously identify common and condition-specific gene modules and quantify their expression levels in each sample in a lower-dimensional space.

Any suggestions on the package are welcome! For technical problems, please report to Issues. For suggestions and comments on the method, please contact Kun ([email protected]) or Vivian ([email protected]).

Installation

You can install scINSIGHT from CRAN with:

install.packages("scINSIGHT")

Usage

Please refer to the package vignette for examples about how to use the package functions.