scMatch

scMatch annotates single-cell RNA sequencing (scRNA-seq) gene expression profiles by matching individual cells to large reference datasets to identify cell types.

Key Features:

• Direct per-cell annotation: Matches each single cell to its closest profile in extensive reference datasets rather than relying on clustering-based averaged expression.
• Low-coverage robustness: Provides reliable annotations for sparse scRNA-seq data produced by low sequencing depth.
• Efficiency and scalability: Implements a rapid matching algorithm that scales to large reference datasets and outperforms SingleR in speed while maintaining comparable accuracy.
• Customizable reference profiles: Supports integration of combined gene expression profiles from multiple sources for tailored reference construction.
• Impact assessment: Evaluates effects of sequencing depth, similarity metrics, and choice of reference datasets on annotation accuracy.
• Implementation and reference: Implemented in Python and compatible with the FANTOM5 reference dataset.

Scientific Applications:

• Large-scale transcriptional analysis: Enables annotation of thousands of single cells within complex tissues for downstream transcriptional studies.
• Precise cell-type identification: Facilitates high-precision identification of cell types to study cellular heterogeneity and function.

Methodology:

Matches individual single-cell profiles against a comprehensive reference dataset using an efficient matching algorithm and evaluates the impacts of sequencing depth, similarity metrics, and reference choice on annotation accuracy.