Scalable nonparametric clustering with unified marker gene selection for single-cell RNA-seq data

Area of Science:

• Genomics
• Computational Biology
• Bioinformatics

Background:

• Single-cell RNA sequencing (scRNA-seq) is crucial for assessing cellular heterogeneity.
• Standard clustering methods often require manual parameter tuning and can lead to high false discovery rates in differential expression analysis.

Purpose of the Study:

• To introduce NCLUSION, a novel nonparametric infinite mixture model for simultaneous clustering and marker gene identification in scRNA-seq data.
• To develop a scalable and statistically robust method for analyzing large scRNA-seq datasets.

Main Methods:

• NCLUSION employs Bayesian sparse priors and a variational inference algorithm.
• The model is designed to handle large-scale scRNA-seq datasets, potentially including millions of cells.

Main Results:

• NCLUSION demonstrates comparable performance to state-of-the-art clustering methods but with significantly reduced computational time.
• The identified clusters are supported by statistically robust and biologically relevant transcriptomic signatures.

Conclusions:

• NCLUSION provides a reliable and efficient tool for hypothesis generation in single-cell biology.
• The method enhances the understanding of expression variation patterns within single-cell populations.