Population-level comparisons of gene regulatory networks modeled on high-throughput single-cell transcriptomics data
Daniel Osorio1, Anna Capasso2, S Gail Eckhardt2
1Department of Oncology, Livestrong Cancer Institutes, Dell Medical School, The University of Texas at Austin, Austin, TX, USA. daniecos@uio.no.
Nature Computational Science
|March 4, 2024
Summary
SCORPION reconstructs comparable gene regulatory networks from single-cell RNA sequencing data, enabling robust population-level comparisons. This tool accurately identifies regulatory differences and advances understanding of disease progression and patient survival.
Area of Science:
- Computational Biology
- Genomics
- Systems Biology
Background:
- Single-cell technologies offer high-resolution insights into molecular mechanisms.
- Data sparsity and cellular heterogeneity pose challenges for modeling biological variability in single-cell studies.
Purpose of the Study:
- To introduce SCORPION, a novel tool for reconstructing comparable gene regulatory networks from single-cell RNA sequencing data.
- To enable accurate population-level comparisons of gene regulatory networks.
Main Methods:
- SCORPION employs a message-passing algorithm to reconstruct gene regulatory networks.
- The tool leverages baseline priors for comparable network reconstruction.
- Performance was evaluated using synthetic data and supervised experiments.
Main Results:
- SCORPION outperformed 12 existing gene regulatory network reconstruction methods on synthetic data.
- The tool accurately identified regulatory network differences between wild-type and perturbed cells.
- SCORPION demonstrated scalability to large datasets, analyzing over 200,000 cells from colorectal cancer tissues.
Conclusions:
- SCORPION provides a robust method for population-level comparison of gene regulatory networks.
- The tool accurately detects differences in tumor regions, consistent across cohorts.
- SCORPION can elucidate phenotypic regulators impacting patient survival in diseases like colorectal cancer.
