rCASC: reproducible classification analysis of single-cell sequencing data.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Single-cell RNA sequencing (scRNA-seq) is crucial for understanding cellular heterogeneity and identifying cell-specific signatures.
• Existing computational tools often lack flexibility and reproducibility in scRNA-seq data analysis.
• The integration of epigenomics, like ATAC-seq, expands scRNA-seq applications, necessitating robust analytical workflows.

Purpose of the Study:

• To develop a flexible and reproducible computational workflow for single-cell RNA sequencing data analysis.
• To provide an integrated environment for analyzing scRNA-seq data, from initial processing to cell subpopulation identification.
• To enhance the reproducibility of scRNA-seq analyses through functional and computational measures.

Main Methods:

• Developed rCASC, a modular workflow utilizing Docker containerization for reproducible analysis.
• Implemented preprocessing tools for quality control and bias reduction (e.g., cell cycle effects).
• Integrated various clustering techniques with novel "cell stability score" (CSS) for robust subpopulation discovery and evaluation.

Main Results:

• rCASC achieves functional and computational reproducibility via Docker containerization.
• The workflow includes tools for cell quality control and bias correction.
• Introduced CSS metric for superior cluster robustness assessment compared to silhouette scores.
• Identified cluster-specific gene signatures effectively.

Conclusions:

• rCASC offers a modular and user-friendly solution for scRNA-seq data analysis.
• Docker ensures ease of installation and computational reproducibility.
• The workflow facilitates the identification of cell subpopulations and their specific markers, enhanced by the CSS metric and a Java GUI for accessibility.