CAGE: Combinatorial Analysis of Gene-cluster Evolution

Abstract

Much important evolutionary activity occurs in gene clusters, where a copy of a gene may be free to acquire new functions. Current computational methods to extract evolutionary information from sequence data for such clusters are suboptimal, in part because accurate sequence data are often lacking in these genomic regions, making existing methods difficult to apply. We describe a new method for reconstructing the recent evolutionary history of gene clusters, and evaluate its performance on both simulated data and actual human gene clusters.

FIG. 1.

An example of duplication inference. (a) The original match regions in a sequence S. S contains 10 matches for regions A_i and where 1 ≤ i ≤ 10, and the box shows the location of an intraposed duplication which is discussed in Sections 2.4 and 2.5. (b) The self-alignment of the original sequence. (c) The self-alignment after rolling back a duplication. (d) The self-alignment after rolling back another duplication. See Section 2.3 for details.

FIG. 2.

An example of the self-alignment change caused by a tandem duplication event.

FIG. 3.

The constraint graphs of matches in Fig. 1a, b. Arcs of type 1, 2, and 3 are represented by dotted, solid, and dashed lines, respectively. In (a), matches m₁ and m₂ have been replaced with m_1* according to the procedure for intraposed duplications discussed in Section 2.2, and node is omitted because it is identical to reappears in (b) after m₃ has been eliminated.

FIG. 4.

An example of an orthologous alignment: (a) Immediately after speciation. (b) After one duplication in species 1.

FIG. 5.

Duplication events inferred for (a) the HLA gene cluster by the combinatorial method, (b) the HLA gene cluster by the MCMC method, (c) the AKR1C gene cluster by the combinatorial method, (d) the AKR1C gene cluster by the MCMC method, and (e) the CYP1A1 gene cluster and (f) the CYP2A gene cluster by the combinatorial method.

FIG. 6.

Simulation results to evaluate detection of duplications. (a) Total numbers of reconstructed events and (b) fraction of true events detected correctly. (c) Simulation results of duplication detection for gene clusters in two species; each bar has three numbers: events before speciation, events in species 1 after speciation, and events in species 2 after speciation, for n true events of each type (i.e., 3n duplications in total) for n = 5, 10, 15, 20, 25. (d) Observed breakpoint reuse rate by the duplications in the human gene clusters.

FIG. 7.

Duplication reconstruction of the DEFA gene cluster in human and orangutan. The orangutan sequence gained 4 more genes from 10 duplications after the split of human and orangutan; the ancestral sequence of the two species had 5 genes. The dotted and shaded parallelograms represent the source and target regions of each duplication, respectively.