Skip to main content
Fig. 1 | BMC Evolutionary Biology

Fig. 1

From: Recurrent sequence evolution after independent gene duplication

Fig. 1

Framework for detecting recurrent sequence evolution between duplicated genes (see text or Material and Methods for details). (0) An alignment of paralog pairs serves as input. (1) The positions in the quartet support different evolutionary scenarios. (2) The two paralog pairs represent two independent duplications if scenario Q is most likely under a model of sequence evolution (LG +Γ10), yielding low D. Only then can we consider patterns in R and S to be recurrent substitutions (as depicted on the tree defined by Q). (3 & bottom left panel) Clustering based on D-scores of all quartets identifies independent duplications in the whole family (two in this case; denoted by red ovals). Here, human and yeast paralogs are classified as independent duplications due to a low weight for scenario Q, unlike human and frog paralogs. (2) For recurrent sequence evolution we require that, after the primary phylogenetic signal (Q), there is a clear secondary signal representing similar fates of paralogs between the two species (F>0). In this case, Human_a is most similar to Yeast_c and Human_b to Yeast_d (nR>nS). (3 & bottom right panel) Clustering based on F-scores of all quartets identifies unique fates in the whole family (two in this case; denoted by yellow and blue ovals). (4) Since yeast paralogs derive from an independent duplication event and are assigned the same two fates as human and frog paralogs, the number of duplications with recurrence (P) is 2. Because the shown alignment is short, the support for recurrence is low \(\left (\overline {Z}_{F}=0.18\right)\)

Back to article page