Efficient learning of association rules from human phenotype ontology

Biomedical ontologies are commonly used to structure and organize formal knowledge about biological and biomedical concepts. Terms structured within ontologies are usually associated to biomedical entities in a process referred to as annotation. The Human Phenotype Ontology (HPO) is a standardized, controlled vocabulary that contains phenotypic information about genes or product genes. Due to the recent introduction of the HPO, problem to check annotation consistency of HPO annotations, has not been formally investigated differently from other ontologies, such as Gene Ontology (GO). In a previous work we introduced a framework to learn association rules from Gene Ontology demonstrating its usefulness to improve annotation consistency. Here we extend those results in HPO and we present a novel framework to learn association rules from HPO. The framework is based on a multithreaded tool able to learn rules in an efficient way. Results demonstrate its usefulness, by extracting rules that connect two or more terms of HPO, currently under investigation.