FastProject

FastProject projects single-cell RNA-Seq gene expression matrices into low-dimensional (two- or three-dimensional) representations and scores cells against annotated gene signatures to characterize phenotypic diversity.

Key Features:

• Input: Processes gene expression matrices derived from single-cell RNA-Seq experiments.
• Low-dimensional projection: Projects high-dimensional gene expression data into two- or three-dimensional representations for visualization and analysis.
• Annotated gene signatures: Incorporates annotated gene sets ("signatures") to contextualize features observed in projections within biological processes.
• Signature scoring: Implements a scoring method that evaluates each cell against a gene signature while minimizing the impact of missed transcripts and variable gene capture rates.
• Signature-projection ranking: Ranks signature–projection pairings to identify associations between projected features and biological processes.
• Modular architecture: Supports integrating and comparing multiple projection methods and gene selection algorithms.

Scientific Applications:

• Phenotypic diversity characterization: Characterizes phenotypic diversity among single cells using combined projection and signature scoring.
• Biological interpretation of projections: Assesses the biological relevance of low-dimensional representations by linking projection features to annotated gene signatures.
• Cellular relationship discovery: Identifies and helps interpret complex cellular relationships and heterogeneity within scRNA-Seq datasets.
• Study of cellular function: Supports investigations into cellular diversity and function using signature-informed projection analyses.

Methodology:

Analyzes single-cell RNA-Seq gene expression matrices, projects data into two- or three-dimensional representations, computes signature scores that minimize effects of missed transcripts and variable gene capture rates, ranks signature–projection pairings, and integrates multiple projection methods and gene selection algorithms.