The effect of branch length variation on the selection of models of molecular evolution.

摘要: Models of sequence evolution play an important role in molecular evolutionary studies. The use inappropriate models may bias the results analysis and lead to erroneous conclusions. Several procedures for selecting best-fit model data at hand have been proposed, like likelihood ratio test (LRT) Akaike (AIC) Bayesian (BIC) information criteria. relative performance these model-selecting algorithms has not yet studied under a range different trees. In this study, influence branch length variation upon selection is characterized. This done by simulating alignments known nucleotide substitution, recording how often true recovered model-fitting strategies. Results study agree with previous simulations suggest that reasonably accurate. However, methods showed distinct levels accuracy. Some LRT approaches better than AIC or BIC Within LRTs, affected complexity initial selected comparisons, only slightly order which parameters are added model. A specific hierarchy starts from simple evolution, performed overall other possible hierarchies, BIC.