pcaReduce

pcaReduce performs agglomerative clustering integrated with principal component analysis (PCA) to identify and hierarchically represent cell states from single-cell transcriptomic data such as single-cell RNA sequencing (scRNA-seq).

Key Features:

• Agglomerative Clustering Approach: Uses an agglomerative clustering method that merges cells based on similarity to build a hierarchical representation of cell states.
• Integration with Principal Component Analysis (PCA): Integrates PCA into the clustering process and associates each branch of the hierarchy with a principal component of variation.
• Comparison and Validation: Validated through comparisons with K-means clustering and traditional hierarchical clustering on real single-cell datasets.
• Enhanced Characterization of Cell States: Links expression data representation and the number of discernible cell types to improve characterization of putative cell states.

Scientific Applications:

• Single Cell Expression Analysis: Applied to scRNA-seq data for identifying and characterizing cellular heterogeneity.
• Cellular Hierarchy Inference: Used to infer cellular hierarchies relevant to developmental biology, disease progression, and tissue organization.
• Complementary Toolset Expansion: Serves as a complementary clustering approach alongside existing single-cell clustering techniques to provide alternative perspectives on heterogeneous cell populations.

Methodology:

Performs agglomerative clustering that merges cells based on similarity while integrating PCA by associating principal components with hierarchy branches, and includes validation via comparison with K-means and traditional hierarchical clustering on real single-cell datasets.