omeClust

omeClust identifies feature clusters in omics datasets by combining measurement similarities with hierarchical data structure and scoring metadata influences to detect biologically relevant communities.

Key Features:

• Simultaneous Incorporation: Combines measurement similarities with the hierarchical structure inherent in omics data to capture relationships among features.
• Resolution Scoring: Evaluates metadata influences on clustering, accommodating varying degrees of similarity within clusters and accounting for both the number of grouped features and their structural organization.
• Performance Superiority: In comparisons using simulated datasets, achieved higher sensitivity and lower misclassification rates when inferring true community structures.

Scientific Applications:

• Diverse omics datasets: Validated on microbial strains, cell line gene expression patterns, and fetal genomic variations.
• Community and functional group discovery: Identifies new communities and functionally related groups within omics data.
• Enrichment scoring and hypothesis generation: Derives enrichment scores linked to biologically meaningful factors to facilitate hypothesis generation.

Methodology:

A clustering mechanism detects feature clusters and scores metadata influences based on their impact on clustering, integrating measurement similarities with the hierarchical data structure to allow variable resolution within clusters.