Characterization and Clustering of GO Terms by Feature Importance Vectors Obtained from Microarray Data

Abstract:- In this paper it is explained a new approach for clustering Gene Ontology (GO) terms by examining microarray data related to them. By segmenting the entire ontology in a single specific level, and applying techniques as discrimination and ranking of features to those GO terms that are contained in that level, it is produced a characterization of the contained terms, as feature importance vectors related to the gene expression patterns that are included in the microarray dataset. By utilizing data mining techniques to cluster the vectors, it is concluded that this new approach may help to obtain relations that are normally hidden among GO terms, not only the ones in the same contained ontology, but also getting a trans-ontological relationship of them