scSorter

scSorter assigns cells to known cell types from single-cell RNA-sequencing (scRNA-seq) data without requiring reference datasets.

Key Features:

• Marker Gene Flexibility: Allows marker genes to be expressed at low levels rather than requiring over-expression for cell-type identification.
• Utilization of Non-Marker Genes: Incorporates expression information from non-marker genes to provide broader cellular expression context for classification.
• No Reference Dataset Requirement: Performs cell type assignment without reliance on external reference datasets for marker calibration or validation.
• High Performance: Demonstrates higher classification power compared to existing methods in empirical evaluations on simulated and real scRNA-seq datasets.

Scientific Applications:

• Single-cell genomics: Assigning cell types and studying cellular heterogeneity from scRNA-seq data.
• Developmental biology: Resolving cell type composition and transitions in developmental studies using scRNA-seq.
• Immunology: Characterizing immune cell populations and states in scRNA-seq datasets.
• Cancer research: Identifying tumor and microenvironment cell types and heterogeneity in scRNA-seq cancer datasets.

Methodology:

Permits low expression of marker genes, incorporates non-marker gene expression, assigns cell types without using reference datasets, and was evaluated on simulated and real scRNA-seq datasets.