Phylogenetic Reconstruction of Orthology, Paralogy, and Conserved Synteny for Dog and Human
Figure 1
The Assignment of Orthology by Ensembl
(A) Shows the true phylogenetic relationships for three dog (D1–3) and three human gene homologues (H1–3). D3 and H3 are 1:1 orthologues, having being derived from a single gene at the last common ancestor (marked “S” for speciation point). D1, D2 and H1, H2 are likewise orthologues of each other but in a many-to-many relationship.
(B) Shows that D1 and H1 and D3 and H3 are BLAST reciprocal best hits (solid arrows; “UBRH” in Ensembl terminology). Because the D2 and H2 loci are closely linked neighbours of the H1 loci, their orthology relationships are also predicted by Ensembl on the basis of their BLAST nonreciprocal best hits: H1 is the best hit for D2, and D2 is the best hit in turn for H2 (dashed red arrows; “RHS” in Ensembl terminology). Because of this lack of reciprocity, H1 is simultaneously in a many-to-one relationship with D2 (and H2) and a one-to-many relationship with D1 and D2. As orthology is, by definition, a transitive property between genes of two species, this inconsistency can be reconciled by linking all four genes together into a single set of orthologues, in effect adding the missing link between D1 and H2. Many such inconsistencies can be found in version 27.1 of the Ensembl Compara database, for example, ENSCAFG00000009718, ENSCAFG00000009724, ENSG00000180305, and ENSG00000182931 are found in relationships illustrated by D1, D2, H1, and H2, respectively.
(C) Human gene H3 has not been predicted. The highest-scoring BLAST alignment for its orphaned orthologue D3 becomes H2 (dashed red arrow). This erroneous assignment of orthology for D3 arises because Ensembl does not distinguish between adjacent in-paralogues such as H1 and H2, and out-paralogues such as H3.