Hierarchical Classification of Gene Ontology with Learning Classifier Systems

The Gene Ontology (GO) project is a major bioinformatics initiative with the aim of standardizing the representation of gene and gene product attributes across species and databases. The classes in GO are hierarchically structured in the form of a directed acyclic graph (DAG), what makes its prediction more complex. This work proposes an adapted Learning Classifier Systems (LCS) in order to predict protein functions described in the GO format. Hence, the proposed approach, called HLCS (Hierarchical Learning Classifier System) builds a global classifier to predict all classes in the application domain and its is expressed as a set of IF-THEN classification rules, which have the advantage of representing more comprehensible knowledge. The HLCS is evaluated in four different ion-channel data sets structured in GO terms and compared with a Ant Colony Optimisation algorithm, named hAnt-Miner. In the tests realized the HLCS outperformed the hAnt-Miner in two out of four data sets.