Gene Ontology friendly biclustering of expression profiles

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Gene expression profiles are used for gene function prediction, assuming similar profiles correlate with similar functions.
• Gene Ontology (GO) organizes gene functions but existing clustering methods fail to reflect GO's hierarchical and complex relationships.
• This limitation reduces confidence in clustering for gene function prediction.

Purpose of the Study:

• To develop a novel clustering technique that preserves the hierarchical relationships found in Gene Ontology (GO).
• To improve the biological relevance and accuracy of gene function prediction through enhanced clustering.
• To address the limitations of current clustering algorithms in reflecting GO's complex category structures.

Main Methods:

• Introduced Smart Hierarchical Tendency Preserving clustering (SHTP-clustering), a bicluster-based method.
• Incorporated Gene Ontology (GO) information directly into the clustering process.
• Developed a Tendency Preserving cluster (TP-Cluster) model to generate a TP-cluster tree.

Main Results:

• The SHTP-clustering algorithm generates a TP-cluster tree where subtrees map effectively to GO hierarchy segments.
• Experiments on yeast cell cycle data demonstrated the method's efficiency and effectiveness.
• The generated TP-Clusters were found to be biologically relevant.

Conclusions:

• SHTP-clustering successfully integrates GO hierarchy into gene expression clustering.
• The method enhances the biological interpretability of gene clusters for function prediction.
• This approach offers a more reliable tool for analyzing gene expression data and predicting gene functions.