NS-Forest

NS-Forest identifies minimal combinations of marker genes necessary and sufficient to classify cell types from single-cell RNA sequencing (scRNA-seq) data.

Key Features:

• Random forest feature selection: Employs random forest feature selection to handle non-linear structure in scRNA-seq data.
• Binary expression scoring: Uses a binary expression scoring approach to select minimal marker gene combinations that discriminate cell types.
• Necessary-and-sufficient marker sets: Outputs minimal marker gene "barcodes" that define semantic cell type identities.

Scientific Applications:

• Cell type identification: Defines cell types by selecting key molecular biomarkers from scRNA-seq data.
• Downstream biological research: Provides marker genes for experimental and mechanistic studies of cellular phenotypes and disease processes.
• Human brain studies: Has been applied to identify marker genes in the human middle temporal gyrus, implicating cell signaling and non-coding RNAs in neuronal identity.

Methodology:

NS-Forest v2.0 applies random forest feature selection together with a binary expression scoring method to scRNA-seq data to identify minimal necessary-and-sufficient marker gene sets.