Discovery of rare cells from voluminous single cell expression data.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Single-cell messenger RNA sequencing (scRNA-seq) enables detailed analysis of cellular transcriptional heterogeneity.
• Droplet-based platforms facilitate high-throughput scRNA-seq, generating massive datasets.
• Identifying rare cell subpopulations within large datasets is computationally challenging for existing algorithms.

Purpose of the Study:

• To develop an efficient algorithm for detecting rare cell subpopulations in ultra-large scRNA-seq datasets.
• To provide a scalable solution for analyzing high-throughput transcriptomic data.
• To aid bioinformaticians in focusing downstream analyses on relevant cell profiles.

Main Methods:

• Introduction of the Finder of Rare Entities (FiRE) algorithm.
• FiRE assigns a rareness score to individual expression profiles.
• Demonstration of FiRE's performance on large-scale scRNA-seq data.

Main Results:

• FiRE processes ultra-large scRNA-seq datasets in seconds.
• The algorithm effectively scores the rareness of individual cell expression profiles.
• Application to mouse brain scRNA-seq data identified a novel pars tuberalis lineage subtype.

Conclusions:

• FiRE offers a computationally efficient method for rare cell detection in large scRNA-seq studies.
• The algorithm facilitates the discovery of novel cell types and subtypes.
• FiRE enhances the analysis of complex biological systems using single-cell transcriptomics.