Multi-Factored Gene-Gene Proximity Measures Exploiting Biological Knowledge Extracted from Gene Ontology: Application...

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Gene Ontology (GO) provides a dynamic vocabulary for describing gene and protein functions.
• GO comprises three sub-ontologies: Biological Process, Cellular Component, and Molecular Function.
• Existing semantic similarity measures often fail to collectively consider GO-term topology and information theory.

Purpose of the Study:

• To propose novel semantic similarity/distance measures for genes based on Gene Ontology (GO).
• To integrate these measures into clustering algorithms for identifying functionally similar genes.
• To compare the performance of proposed measures and algorithms against existing methods.

Main Methods:

• Developed three novel semantic similarity/distance measures incorporating diverse aspects of the GO-tree.
• Implemented these measures within multi-objective and single-objective clustering frameworks.
• Conducted comparative analysis using 10 popular existing GO-based measures and tools.

Main Results:

• Experimental results on Mouse, Yeast, and Human genome datasets show the superiority of multi-objective clustering with the proposed measures.
• The novel measures effectively capture different topological and information theoretic aspects of GO-terms.
• Clustering outcomes were validated through biological and statistical significance tests.

Conclusions:

• The proposed multi-factored semantic similarity measures enhance gene clustering accuracy.
• Multi-objective clustering algorithms combined with these novel measures offer a superior approach for functional gene identification.
• This work contributes to improved bioinformatics analysis for gene function and disease association studies.