Prediction of homology and divergence in the secondary structure of polypeptides.

Publisher Website
Google Scholar
Abstract
A quantitative procedure is described for the comparison of secondary structure of homologous proteins. Standard predictive methods are used to generate probability profiles from pairs of homologous amino acid sequences; correlation coefficients (R) are then computed between each pair of amino acids for .alpha.-helix (R.alpha.), extended structure (R.beta.), turn (Rt), and coil (Rc). R values are > 0.2 for correctly aligned homologous sequences. Unrelated or incorrectly aligned sequences give R values near zero. Lack of correlation for a segment of otherwise well-correlated sequences is used to identify structural divergence, which is then evaluated graphically by using difference profiles. A combination of these techniques correctly predicts secondary structural differences between melittin or .beta.-endorphin and their respective synthetic analogs. The method is potentially useful to describe evolutionary changes in protein secondary structure as well as in the design of peptide analogs.