engleski

A method for functional classification of globular proteins

Determining function of a protein from its primary structure is among the most important problems in molecular biology. The usual approach – inferring function from homology – requires detailed description of functional units, considerable amount of data and is computationally rather demanding. Instead, we describe here a simple method for dividing a family of proteins into two functional subfamilies that is based on an underlying model of evolution. More precisely, we consider the alignment of a protein family and choose the sites in the alignment that strongly prefer bimodal evolutionary model over the unimodal one. In order to obtain two functional groups, we combine the information from these sites by an unsupervised learning algorithm. The method is demonstrated by analysing families of AT and KR domains from type I polyketide synthases.

Protein function; primary structure; inferring function from homology; protein family; alignment; bimodal evolutionary model; AT and KR domains

nije evidentirano