Genome-wide assessment of ortholog quality in Ensembl

Orthologs prediction can be a difficult exercise, particularly on a large scale. While methods to cluster genes into protein families have improved greatly with the use of HMM profiling, verification of orthologs should be performed for robustness. Here, we propose a new quality check that focuses on large-scale rearrangement information. Gene neighbourhoods surrounding each gene of a pair of orthologs rightly are expected to be well conserved. Our approach makes the assumption that rearrangements are likely to happen to a group of contiguous genes, rather than genes in isolation (1). Our method uses local synteny information to calculate a percentage score for orthologous pairs of genes (2 genes upstream and 2 genes down stream of each gene of the queried pair). The higher the gene order conservation (GOC) percentage score of an ortholog the more confidence we have in the prediction of said ortholog. The distribution of the GOC percentage score for all the orthologs inferred for a genome can also be used as an indication of the quality of the assembly of that genome. We have also used the distribution of the GOC percentage score for all the orthologs inferred for a set of species to compare two gene tree pipelines employing 2 different approaches to gene tree inference. Our GOC scores are used in combination with whole-genome-alignments-based (WGA) scores and more traditional filters such as a threshold on the percentage of sequence identity to identify a subset of high-confidence orthologs. The GOC and WGA scores and the sets of high-confidence orthologs are available since version 86 of Ensembl.