scClassify

scClassify performs multiscale classification of single-cell RNA-seq data using ensemble learning and cell type hierarchies to enable sample size estimation for accurate cell type classification and joint classification with multiple references.

Key Features:

• Ensemble Learning Classification: scClassify employs ensemble learning methods for cell type classification across diverse scRNA-seq datasets.
• Cell Type Hierarchy Construction: scClassify builds hierarchical structures of cell types from single or multiple annotated reference datasets.
• Sample Size Estimation: scClassify estimates the sample size required for accurate classification of cell types within a cell type hierarchy.
• Joint Classification with Multiple References: scClassify enables joint classification of cells when multiple reference datasets are available.
• Scalability to Large Atlases: scClassify scales to large single-cell atlases for automated cell type identification.

Scientific Applications:

• Automated Cell Type Identification: scClassify identifies cell types in scRNA-seq data using well-annotated reference datasets.
• Cross-Technology Classification: scClassify classifies cells across diverse combinations of dataset sizes, sequencing technologies, and complexity levels.
• Novel Subpopulation Discovery: scClassify identifies previously unidentified subpopulations of cells in large-scale datasets such as Tabula Muris.

Methodology:

scClassify uses ensemble learning combined with cell type hierarchies constructed from annotated reference datasets, with validation across 114 pairs of reference and testing data and demonstration through simulations and compendia of experimental datasets.