Improved Biomolecular Annotation Prediction through Weighting Scheme Methods

Biomolecular annotation databases are very important in modern biomedical and genetic research. Correct interpretation of biological experiments depends on consistency and completeness of such databases. To improve their quality and coverage, computational methods that are able to supply a ranked list of predicted gene or gene products annotations are extremely useful. In this paper we propose an improvement of two of these annotation prediction methods, which are based on vectorial semantic analysis. Such improvement relies on a preparatory weighting of the set of considered annotations. We tested the efficacy of the weighting approach on the Gene Ontology annotations of the genes of three organisms. Evaluation results clearly demonstrated effectiveness of the weighting approach, which considerably enhances the results of the considered annotation prediction methods.