Pseudo-Rate Matrices, Beyond Dayhoff’s Model

One of the fundamental techniques biology is sequence alignment, namely transforming one into another with minimal change. Sequence alignment essential for evolutionary studies and a source information analysis physico-chemical mechanisms which are at heart protein activity. Biologists almost exclusively use methods based on very simple model, although they aware that this can be quite removed from reality. In fact, more complex models involve so many variables cannot calculated in practice. This paper presents method to estimate quality approximation made using models, giving measure deviation It pairwise alignments, without resorting multiple therefore requiring construction trees problems associated it. The also describes an approach allows building clusters sequences strongly relying choice dissimilarity measure. illustrates interest effectiveness point view promoted by Alex: assume as little possible try gather data, before turning explicit modeling if necessary.