Skip to main content
Figure 1 | BMC Evolutionary Biology

Figure 1

From: Efficient context-dependent model building based on clustering posterior distributions for non-coding sequences

Figure 1

Model-switch integration for the Ancestral Repeats dataset. Results for the two model-switch integration schemes: (a) annealing, i.e. β increases from 0 (independent GTR model) to 1 (GTR16C full context-dependent model) and (b) melting, i.e. β decreases from 1 to 0. The comparison between a single run (left) and a composite run using ten intervals (right) reveals almost identical log Bayes Factor estimates. The composite run yields a slightly broader confidence interval around the log Bayes Factor estimate.

Back to article page