An Approach for the Alignment of Biomedical Ontologies based on Foundational Ontologies

Abstract

Genome annotation, the task of assigning a description to each discovered genome sequence, is an important activity within the process of sequencing. It relies on the use of ontologies to maintain a uniform vocabulary and to support interoperability of different information resources. Often, only the Gene Ontology (GO) is used in the annotation process, but the exploration of other ontologies, along with GO, could enrich the vocabulary used in the annotation, complementing it with more details. To facilitate this task, it is necessary the identification of equivalences between terms from GO and terms from other ontologies. This work presents an approach for aligning biomedical ontologies within the genome annotation process that helps to identify equivalent terms between GO and other biomedical ontologies, enabling the annotator to choose which among them is more suitable as a descriptor. Two main points have guided its development: (i) the choice of a subset of similarity measures suited to the characteristics of biomedical ontologies; and (ii) the use of foundational ontologies, allowing the analysis of the conceptual nature of each term, which serves as a fundamental parameter for similarity calculation, reducing the possibility of associations between terms derived
from different categories. Initial experiments showed an improvement on the alignment quality, represented by a 14% increase in the number of correct alignments and a 5%  decrease in the number of incorrect associations, reinforcing the usefulness of the proposed approach in supporting the annotator's work.