Enigma

Enigma extracts gene expression modules from perturbational microarray datasets to identify partially coexpressed gene groups arising from chemical and genetic perturbations.

Key Features:

• Differential Expression Integration: Incorporates differential expression analysis results (unlike traditional biclustering methods) to guide module extraction from perturbational datasets.
• Reduction of Redundancy: Automatically optimizes core clustering parameters to minimize redundancy between extracted modules.
• Internal Substructure Recognition: Identifies internal substructures in modules where subsets of genes exhibit distinct yet significantly related expression patterns.
• Handling Overlapping Clusters: Generates and evaluates overlapping clusters using a quality criterion tailored to account for redundancy.

Scientific Applications:

• Systems biology: Infers gene networks and regulatory mechanisms underlying cellular functions.
• Perturbational expression compendia analysis: Analyzes perturbational expression compendia to reveal relationships induced by chemical and genetic perturbations.
• Method benchmarking: Demonstrated superior performance on artificial datasets using a quality criterion that accommodates overlapping clusters and redundancy.
• Yeast expression studies: Applied to the Rosetta compendium of Saccharomyces cerevisiae expression profiles to generate detailed biological predictions.

Methodology:

Employs combinatorial statistics and graph-based clustering, integrates differential expression analysis results, automatically optimizes core clustering parameters to reduce redundancy, uses a quality criterion for overlapping clusters, and recognizes internal substructures within modules.