Constructing a rate matrix for amino acid evolution from pairs of sequences.

摘要

The substitution rate matrices are necessary to model evolution of taxa based on amino acid sequences. We develop an estimation method which is free of some of the limitation of the counting and maximum likelihood methods, and still gives us the reasonable estimates. We build on Arvestad and Bruno (1997) and develop a method constructing the rate matrix from the dataset with multiple protein families without any limitation on data selection, no topology construction and no iteration. Our testing results of a DNA dataset and a multiple protein families dataset selected from PANDIT dataset showed that our method is reasonable.