Regular Talk:
Computer simulation replacements aminoacids in proteins

Abstract

We constructed logical-probabilistic models determined structure
preserving mutations in proteins. For model generation we used
Neural Networks and modified method KRAB. As a data source for the
algorithms we used physical-chemical and others proteins'
characteristics as well as proteins free energy's changing under
mutation. Algorithms were implemented. Designed software is able
to predict allowable protein's mutations. In the paper it is also
presented a comparison between the used methods. It is shown that
algorithm KRAB makes more accurate predictions with respect to
used data source.