CHAI

CHAI aggregates clustering results from multiple scRNAseq clustering methods to identify cell types through similarity matrix integration.

Key Features:

• CHAI-AvgSim: Aggregates clustering results from multiple scRNAseq clustering methods through average similarity matrix integration.
• CHAI-SNF: Aggregates clustering results from multiple scRNAseq clustering methods through similarity network fusion.
• Extensible framework: Provides a framework for incorporating new clustering algorithms into the aggregation pipeline.
• Multiomic integration: Supports integration of additional data types including spatial transcriptomics data.

Scientific Applications:

• Cell-type identification: Identifies cell types from scRNAseq data by aggregating results from seven state-of-the-art clustering methods.
• Tumor cell cluster identification: Identifies leader tumor cell clusters enriched with specific markers such as CDH3.
• Spatial transcriptomics analysis: Improves clustering performance by including spatial transcriptomics data through CHAI-SNF.

Methodology:

CHAI implements two competing methods for consensus clustering through similarity matrix integration: CHAI-AvgSim computes average similarity across clustering method outputs, and CHAI-SNF applies similarity network fusion. Both methods aggregate clustering results from seven state-of-the-art scRNAseq clustering methods.