EDISA

EDISA identifies gene expression modules in three-dimensional gene-condition-time datasets using a probabilistic iterative clustering approach to characterize co-regulated genes across conditions and time.

Key Features:

• Probabilistic Clustering Approach: Samples initial gene expression modules from three-dimensional gene-condition-time datasets using a probabilistic framework grounded in mathematical definitions of gene expression modules.
• Iterative Refinement Process: Refines sampled modules by iteratively removing genes and conditions that do not meet predefined module criteria.
• Extension for Maximality: Extends refined modules to ensure maximal inclusion of genes and conditions.
• Mathematical Module Definitions: Employs explicit mathematical definitions of gene expression modules to guide sampling and refinement.
• Application to Synthetic and Real Microarray Datasets: Applied to synthetic datasets—recovering implanted modules despite varying levels of background noise—and to real microarray datasets.

Scientific Applications:

• Independent Response Profile Modules: Identifies modules of genes co-regulated across multiple conditions with varying response patterns under each condition.
• Coherent Modules: Detects genes showing similar responses across all conditions, frequently occurring within independent response profile modules.
• Condition-Specific Response Modules: Captures gene expression modules specific to a single condition.
• Organization of Transcriptional Control: Reveals global patterns of co-regulation to characterize the organization of transcriptional control across conditions and time.

Methodology:

Integrates multiple time-series datasets into a unified three-dimensional gene-condition-time dataset, samples initial modules probabilistically using mathematical definitions of gene expression modules, iteratively refines modules by removing noncompliant genes and conditions, and extends modules to ensure maximality.