PASCCA

PASCCA performs canonical correlation analysis-based clustering and analysis of alternative polyadenylation (APA)-related gene expression by integrating poly(A) site and gene-level data from 3' end sequencing.

Key Features:

• Characterization of Poly(A) Sites: Characterizes poly(A) sites by abundance and relative usage using 3' end deep sequencing-derived poly(A) site data.
• Quantification of Gene Associations: Quantifies associations between genes by modeling poly(A) site information within genes and measuring associations among poly(A) sites across different genes.
• CCA-based Clustering: Clusters APA-related genes using canonical correlation analysis (CCA) while integrating multiple layers of information from poly(A) site and gene levels and accounting for replicate number and within-group variability.
• Performance Evaluation: Evaluates clustering on real and synthetic poly(A) site datasets using connectivity, Dunn index, average distance, average distance between means, and biological homogeneity index, reporting improved metrics relative to traditional distance measures.

Scientific Applications:

• Inference of APA-specific Gene Modules: Applied to published poly(A) site data of rice to infer APA-specific gene modules and reveal distinct functional gene modules.
• Analysis of APA Dynamics Across Conditions: Analysis of APA across biological conditions using 3' end sequencing to elucidate interactions between genes and their APA sites.

Methodology:

Implements canonical correlation analysis (CCA) to cluster genes, integrates poly(A) site- and gene-level data, characterizes poly(A) sites by abundance and relative usage from 3' end deep sequencing, quantifies associations among poly(A) sites and genes, accounts for replicate number and within-group variability, and evaluates clustering performance on real and synthetic poly(A) site datasets using connectivity, Dunn index, average distance, average distance between means, and biological homogeneity index.