On entropy and information in gene interaction networks

Area of Science:

• Bioinformatics
• Systems Biology
• Computational Biology

Background:

• Biological experiments yield gene lists requiring assessment of their biological relevance and interrelationships.
• Existing gene enrichment tests primarily rely on gene identity, limiting the assessment of functional relatedness within interaction networks.

Purpose of the Study:

• To develop novel methods for quantifying relatedness between gene sets within biological interaction networks.
• To extend the concept of gene set relatedness beyond simple gene overlap, incorporating network topology and information theory.

Main Methods:

• Derivation of entropy, interaction information, and mutual information for gene sets based on interaction networks.
• Application of a phenomenological model of a living cell, treating interacting genes as harmonic oscillators.
• Exact calculation of information quantities due to the quadratic nature of the energy function.

Main Results:

• The developed methods accurately estimate probabilities of mutual information between gene sets from independent genome-wide association studies and metabolic pathways.
• Novel relationships among human diseases, inferred from gene signatures, were identified and validated observationally.
• The approach successfully detected relationships not discernible by conventional gene enrichment methods.

Conclusions:

• The proposed information-theoretic measures provide a powerful, complementary approach to gene enrichment for understanding gene set relationships.
• These methods enhance the interpretation of complex biological data by capturing network-based functional associations.
• The R package 'gsia' is available for implementing these novel gene set analysis techniques.