scConsensus

scConsensus integrates supervised and unsupervised clustering in an R-based framework to improve cell type identification and annotation in single-cell RNA sequencing (RNA-seq) datasets.

Key Features:

• Integration of Clustering Approaches: Combines unsupervised clustering that identifies clusters from inherent data patterns with supervised clustering that uses a reference panel of labeled transcriptomes to guide cluster formation and cell type identification.
• Consensus Clustering Generation: Generates consensus clusters by integrating results from both unsupervised and supervised approaches to reconcile complementary insights.
• Refinement Using Differentially Expressed Genes (DEGs): Refines consensus clusters using differentially expressed genes (DEGs) to enhance specificity and biological relevance of cell type annotations.
• Application to Real-World Data: Applies the framework to existing single-cell RNA-seq datasets, including sorted peripheral blood mononuclear cell (PBMC) sub-populations.

Scientific Applications:

• Single-cell transcriptomics: Improves precision of cell type identification and annotation in single-cell RNA-seq studies.
• Immunology: Enables analysis of immune cell heterogeneity, including characterization of PBMC sub-populations.
• Developmental biology: Facilitates identification of distinct cell states during development through refined clustering and DEG-based annotation.
• Disease pathology: Supports investigation of cellular heterogeneity and disease-associated cell types in pathological samples.

Methodology:

Performs unsupervised and supervised clustering on single-cell RNA-seq data, integrates the clustering results into consensus clusters, and refines clusters using differentially expressed genes (DEGs).