p.p1 {margin: 0.0px 0.0px 0.0px 0.0px; font: 6.5px Helvetica}\n

Partial homology to a hybrid (fusion) protein causes distinct orthologous groups to overlap. (a)The five proteins involved in overlapping [labeled (a-e)] are

\n

depicted as rectangles and grouped together into two overlapping groups (a,b,c and c,d,e), where protein c is the hybrid having partial homology to both groups. (b) The

\n

protein similarity graph of significant similarities between the proteins. Two phylogenetically unrelated protein groups are joined together. (c) Diagram illustrating how

\n

different databases handle the grouping of these proteins: (i) KOG (K); (ii) InParanoid (I); and (iii) HomoloGene (H), OrthoMCL-DB (O). In the current example, only the KOG

\n

database reflects the orthologous relationships between the proteins correctly, leading to a reliable inference of the protein phylogenies (iv). It should be emphasized that a

\n

phylogenetic gene tree cannot be constructed from the protein similarity graph in panel (b), because this group includes proteins that have no mutual sequence similarity at

\n

all [(a,b) versus (d,e)]. Graph nodes correspond to UniProt accessions: a, dihydrofolate reductase of Drosophila melanogaster (fruit fly), P17719; b, dihydrofolate reductase

\n

of Homo sapiens (human), P00374; c, bifunctional dihydrofolate reductase-thymidylate synthase 2 of Arabidopsis thaliana (thale cress), Q05762; d, thymidylate synthase of

\n

fruit fly, O76511; e, thymidylate synthase of human, P04818.

\n

Orthology