DEEP

DEEP predicts key effectors from differential gene expression by integrating high-throughput transcript abundance data with biomolecular interaction networks such as TRANSPATH.

Key Features:

• Integration of Gene Expression Data: Accepts high-throughput transcript abundance measured by Serial Analysis of Gene Expression (SAGE) or microarrays across tissues, conditions (e.g., normal vs. malignant), or time points.
• Incorporation of Biological Expert Knowledge: Integrates interaction and signal transduction pathway information from the TRANSPATH database to inform network-based analysis.
• Graph-Based Analysis: Assigns significance values to genes based on over-expression and constructs a graph where nodes represent signaling components and edges denote signaling events.
• Graph Traversal and Significance Inheritance: Traverses the graph and propagates inherited significance values from starting nodes to encountered nodes to identify implicated molecules that are not differentially expressed.
• Visualization: Computes weighted-average inherited significance per node and visualizes the graph with node colors ranging from green (significant for tissue/condition A) to yellow (not significant) to red (significant for tissue/condition B).

Scientific Applications:

• Biomarker identification: Highlights candidate biomarkers by linking differential transcript abundance to network-positioned effectors.
• Mechanism elucidation: Reveals molecular components and signaling events potentially underlying phenotypic differences between tissues or conditions.
• Hypothesis generation: Produces network-informed predictions of additional genes or gene products for experimental validation.

Methodology:

Integrates SAGE or microarray transcript abundance with TRANSPATH interaction data, assigns significance values based on gene over-expression, constructs a graph of signaling components (nodes) and signaling events (edges), traverses the graph propagating significance to encountered nodes, and computes weighted-average inherited significance per node for color-coded visualization.