2000 , Volume 5, Special issue, p.79-86

Analysis of protein physico-chemical characteristics brings about the problem of estimation of their distribution parameters (mean, variance, and covariance) over a sample of sequences of a protein family. Evolutionary relationship between the proteins analyzed result in inefficiency of the estimations made under the assumption that they are statistically independent. In this work, we propose a method based on the model of evolution of protein physico-chemical characteristics as a Brownian process for taking into account the evolutionary relationship of proteins. Efficiency of this approach was tested using the data obtained through numerical simulation of protein family evolution. We have demonstrated that the approach proposed gives better estimates of distribution parameters than the estimates made under the assumption of statistical independence of the sequences. The approach also allows the rate of type I error of the estimates of the correlation coefficients between the amino acid characteristics at the sequence positions to be decreased.