scMRMA: single cell multiresolution marker-based annotation.

Area of Science:

• Genomics
• Computational Biology
• Immunology

Background:

• Single-cell RNA sequencing (scRNA-seq) is crucial for analyzing cellular heterogeneity.
• Accurate cell identity determination is essential for interpreting scRNA-seq data.
• Current methods often use unidirectional projection, limiting annotation resolution.

Purpose of the Study:

• To develop a novel bidirectional cell annotation method, scMRMA.
• To improve the accuracy and resolution of cell identity prediction in scRNA-seq data.
• To apply scMRMA for biological discovery in cancer immunology.

Main Methods:

• Developed scMRMA, a bidirectional annotation tool utilizing a hierarchical reference.
• Implemented iterative clustering and deep annotation strategies.
• Benchmarked scMRMA against existing methods on four independent datasets.

Main Results:

• scMRMA demonstrated superior performance compared to existing methods in cell annotation accuracy.
• The method achieved enhanced resolution in identifying cell populations.
• Successfully identified CD8 T cell expansion in squamous cell carcinoma post-anti-PD-1 therapy.

Conclusions:

• scMRMA offers a more accurate and higher-resolution approach for cell identity determination.
• The bidirectional strategy effectively leverages hierarchical references for improved annotation.
• scMRMA provides valuable insights into immune cell dynamics in cancer treatment.