A statistical method for high-throughput screening of predicted orthologs

Orthologs are genes in different species that diverged from a common ancestral gene after speciation. Their identification is critical for reliable prediction of gene function in newly sequenced genomes. Orthologous genes are usually identified by a high-throughput method called Reciprocal-Best BLAST-hit (RBH). As RBH is subject to errors from incomplete sequencing or gene loss in a species, a bioinformatics tool called Ortholuge was developed that identifies RBH-predicted orthologs with atypical genetic divergence. However, declaring the cut-off for atypical divergence in Ortholuge is very computationally-intensive, and so we propose a faster statistical procedure and examine its performance by simulation. We find that performance depends on the fit of the assumed model for the distribution of divergence measures in true orthologs.