Orthology detection still requires more effective scaling algorithms. Combinations of alignment, synteny, evolutionary distances and protein interactions have been used in different unsupervised algorithms to improve effectiveness while many available databases are concerned with the scaling problem. In this paper, a set of gene pair features based on similarity measures, such as alignment scores, sequence length, gene membership to conserved regions and physicochemical profiles are combined in a supervised Pairwise Ortholog Detection (POD) approach to improve effectiveness considering low ortholog ratios in relation to all possible pairwise comparisons between two genomes. In this POD scenario, big data supervised classifiers managing imbalance between ortholog and non-ortholog pair classes allow for an effective scaling solution built from two genomes and extended to other genome pairs.

The supervised approach for POD was compared with Reciprocal Best Hits (RBH), Reciprocal Smallest Distance (RSD) and a Comprehensive, Automated Project for the Identification of Orthologs from Complete Genome Data (OMA) algorithms by using (i) Saccharomyces cerevisiae - Kluyveromcyes lactis, (ii) Saccharomyces cerevisiae - Candida glabrata and (iii) Saccharomyces cerevisiae - Schizosaccharomyces pombe yeast genome pairs as benchmark datasets. Four datasets derived from each genome pair comparison with different alignment settings were used. Because of the large amount of instances (gene pairs) and the data imbalance, the building and testing of the supervised model was only possible by using big data supervised classifiers managing imbalance. Evaluation metrics taking low ortholog ratios into account were applied. From the effectiveness perspective, MapReduce Random Oversampling combined with Spark Support Vector Machines outperformed RBH, RSD and OMA, probably, because of the consideration of gene pair features beyond alignment similarities combined with the advances in big data supervised classification.