Improving the study of plant evolution with multi-matrix mixture models

Abstract

Amino acid substitution model is a key component to study the plant evolution from protein sequences. Although single-matrix amino acid substitution models have been estimated for plants (i.e., Q.plant and NQ.plant), they are not able to describe the rate heterogeneity among sites. A number of multi-matrix mixture models have been proposed to handle the site-rate heterogeneity; however, none are specifically estimated for plants. To enhance the study of plant evolution, we estimated both time-reversible and time non-reversible multi-matrix mixture models QPlant.mix and nQPlant.mix from the plant genomes. Experiments showed that the new mixture models were much better than the existing models for plant alignments. We recommend researchers to use the new mixture models for studying the plant evolution.