SifiNet: a robust and accurate method to identify feature gene sets and annotate cells.

Area of Science:

• Computational Biology
• Genomics
• Bioinformatics

Background:

• Accurate identification of cellular subpopulations is crucial for understanding complex biological systems.
• Existing single-cell analysis pipelines often rely on clustering, which can introduce inaccuracies.
• Integrating multi-omic data (e.g., single-cell RNA and ATAC sequencing) offers a more comprehensive view of cellular states.

Purpose of the Study:

• To develop a robust computational pipeline, SifiNet, for precise identification and annotation of cellular subpopulations.
• To elucidate intrinsic relationships among cellular subpopulations without relying on a cell clustering step.
• To enable comprehensive multi-omic cellular profiling using single-cell RNA and ATAC sequencing data.

Main Methods:

• SifiNet is a computational pipeline designed for gene set identification and cellular subpopulation analysis.
• It uniquely bypasses the traditional cell clustering stage common in other annotation pipelines.
• The pipeline integrates and analyzes both single-cell RNA sequencing (scRNA-seq) and single-cell ATAC sequencing (scATAC-seq) data.

Main Results:

• SifiNet demonstrates superior performance compared to state-of-the-art methods across multiple experimental datasets.
• The pipeline accurately identifies distinct gene sets and annotates cellular subpopulations.
• It effectively elucidates intrinsic relationships among these identified subpopulations.
• SifiNet provides comprehensive multi-omic cellular profiles by analyzing both scRNA-seq and scATAC-seq data.

Conclusions:

• SifiNet offers a robust and accurate alternative for single-cell multi-omic data analysis.
• By circumventing the cell clustering stage, SifiNet mitigates potential inaccuracies, leading to more reliable results.
• The open-source R package facilitates broader adoption and application in biological research.