Interpretable single-cell factor decomposition using sciRED.

Area of Science:

• Genomics
• Computational Biology
• Bioinformatics

Background:

• Single-cell RNA sequencing (scRNA-seq) reveals cellular heterogeneity but faces challenges from technical noise, sparsity, and high dimensionality.
• Existing data factorization methods require manual interpretation of complex biological signals.
• Improved interpretability is crucial for extracting meaningful biological insights from scRNA-seq data.

Purpose of the Study:

• To develop a novel computational method, Single-Cell Interpretable REsidual Decomposition (sciRED), for enhanced interpretation of scRNA-seq factor analysis.
• To address limitations in current methods by improving the identification and characterization of biological signals within scRNA-seq data.

Main Methods:

• sciRED preprocesses scRNA-seq data by removing known confounding effects.
• It employs factor rotation techniques to enhance interpretability and maps factors to known covariates.
• The method identifies unexplained factors for potential hidden biological phenomena and determines associated genes and biological processes.

Main Results:

• Application of sciRED to diverse scRNA-seq datasets revealed sex-specific variations in kidney data.
• It successfully discerned varying immune stimulation signals in peripheral blood mononuclear cell (PBMC) data.
• sciRED aided in reducing ambient RNA contamination in liver data, facilitating strain variation analysis and identifying rare cell types and zonation programs.

Conclusions:

• sciRED significantly improves the interpretability of factor analysis in scRNA-seq data.
• The method is effective in characterizing a wide range of biological signals, including technical and biological variations.
• sciRED offers a valuable tool for deeper biological discovery from complex single-cell transcriptomic datasets.