Related Experiment Video
Updated: Jul 12, 2025
06:24
Multiplexed Analysis of Retinal Gene Expression and Chromatin Accessibility Using scRNA-Seq and scATAC-Seq
Published on: March 12, 2021
3.6K
DeMixSC: a deconvolution framework that uses single-cell sequencing plus a small benchmark dataset for improved
Shuai Guo1,2, Xiaoqian Liu1,2, Xuesen Cheng3,2
1Department of Bioinformatics and Computational Biology, The University of Texas MD Anderson Cancer Center, Houston, TX, USA.
Biorxiv : the Preprint Server for Biology
|October 24, 2023
Summary
Technological differences in sequencing data hinder accurate cell type deconvolution. We developed DeMixSC, a framework using matched data to improve deconvolution accuracy for complex biological samples.
Area of Science:
- Genomics
- Bioinformatics
Background:
- Bulk deconvolution using single-cell/nucleus RNA-seq (sc/nRNA-seq) is vital for analyzing biological sample heterogeneity.
- Technological discrepancies between sequencing platforms reduce deconvolution accuracy.
Approach:
- Introduced an experimental design to match inter-platform biological signals, identifying technological discrepancies.
- Developed DeMixSC, a deconvolution framework based on weighted nonnegative least-squares, utilizing matched benchmark data.
- DeMixSC adjusts for genes with high technological discrepancy and aligns benchmark data with large patient cohorts for scalable deconvolution.
Key Points:
- DeMixSC significantly improves deconvolution accuracy, as demonstrated with a healthy retina benchmark dataset.
- The framework shows broad applicability, validated on a cohort of 453 age-related macular degeneration patients.
- Highlighted the critical impact of technological discrepancy and the necessity of well-matched datasets.
Conclusions:
- DeMixSC offers a robust solution for accurate bulk deconvolution in large patient cohorts.
- The framework is broadly applicable to disease tissues, including potential applications in cancer research.
