CIPR

CIPR predicts the biological identity of cell clusters from single-cell RNA sequencing (scRNAseq) experiments by scoring cluster-level gene expression profiles against reference datasets to assign cell-type identities.

Key Features:

• Reference-based scoring: Rapidly scores unknown cluster gene expression against established mouse (2) and human (5) reference datasets and user-supplied custom reference datasets.
• Inter-species comparison: Supports comparisons between species when assigning cluster identities.
• Multiple scoring methodologies: Computes identity scores at the cluster level using multiple, alternative scoring approaches.
• Gene filtering: Allows filtering of lowly variable genes to improve discriminatory power.
• Reference subset exclusion: Enables exclusion of irrelevant reference cell subsets to focus comparisons.
• Input compatibility: Accepts inputs generated by popular scRNAseq analysis software for cluster-level annotation.
• R package implementation: Provided as an R package for integration into computational workflows.
• Performance benchmarking: Demonstrated to be less computationally intensive and faster than comparable software while maintaining high accuracy, including in analyses of tumor-infiltrating immune cells.

Scientific Applications:

• Cell cluster annotation: Assigns putative cell-type identities to clusters derived from scRNAseq experiments.
• Tumor immunology: Identifies tumor-infiltrating immune cell populations from scRNAseq data.
• Cross-species studies: Facilitates inter-species comparisons of cell-type identities.
• Custom-reference analyses: Enables annotation using specialized or study-specific reference datasets.

Methodology:

Scores cluster-level gene expression against reference gene expression profiles using multiple identity-scoring methods, with optional filtering of lowly variable genes, exclusion of reference subsets, support for inter-species comparisons and custom references, and input intake from common scRNAseq analysis outputs.