CAMML

CAMML assigns multi-label cell-type scores to single-cell RNA-sequencing (scRNA-seq) data and integrates CITE-seq protein marker measurements to improve cell-type classification using Variance Adjusted Mahalanobis Distances.

Key Features:

• Multi-Label Cell Typing: Assigns multiple potential cell-type labels per cell by scoring against cell-type-specific weighted gene sets.
• Variance Adjusted Mahalanobis Distances (VAM): Uses VAM to adjust for variance in gene expression data and improve robustness of cell-type scoring.
• Integration with Multi-Omic Data: Integrates transcriptomic scRNA-seq data with proteomic measurements from CITE-seq to combine complementary information.
• ChIMP (Integration of Marker Proteins): Refines CAMML scores by incorporating discretized cell-type marker protein measurements to produce more conservative multi-omic cell-type assignments.

Scientific Applications:

• Cell-type identification in diseased tissues: Enables accurate annotation of cell types in complex and disease-affected tissues using scRNA-seq and CITE-seq data.
• Studies of cellular heterogeneity and molecular underpinnings of disease: Facilitates analysis of cellular composition and multi-omic signatures underlying disease processes.

Methodology:

Scores cells using weighted gene sets, applies Variance Adjusted Mahalanobis Distances (VAM) to adjust for expression variance, and for ChIMP discretizes CITE-seq protein marker measurements and integrates them with scRNA-seq-derived cell-type scores.