WMAXC

WMAXC identifies condition-specific sub-networks by integrating gene expression and protein-protein interaction data and optimizing a weighted maximum-clique objective to select connected gene sets relevant to specific diseases or cellular conditions.

Key Features:

• Integration of Data Types: Integrates gene expression data with protein-protein interaction information to construct molecular networks.
• Condition-Specific Analysis: Identifies sub-networks that are specific to particular conditions, cell states, or disease environments.
• Scoring Functions: Employs scoring functions that jointly assess condition-specific changes in individual gene expression and gene-gene co-expression.
• Optimization Approach: Formulates a modified maximum clique problem optimizing a quadratic objective under sparsity constraints to select dense relevant sub-networks.
• Algorithmic Strategy: Combines a continuous genetic algorithm with an efficient projection procedure to maximize the scoring function while enforcing sparsity.
• Graph Representation: Represents genes as nodes and interactions or co-expression strengths as weighted edges in a graph model.

Scientific Applications:

• Disease Research (ovarian and prostate cancer): Selects genes from ovarian and prostate cancer datasets that are enriched in pathways relevant to cancer progression.
• Pathway Enrichment Analysis: Captures condition-pertinent gene subsets to support pathway enrichment and interpretation of disease mechanisms.

Methodology:

Constructs a weighted graph with genes as nodes and edges encoding interaction or co-expression strength; applies scoring functions that evaluate condition-specific changes in expression and interactions; solves an optimization problem akin to finding a maximum scored clique by optimizing a quadratic objective under sparsity constraints using a continuous genetic algorithm combined with an efficient projection procedure.