Predictive estimation of protein linear epitopes by using the program PEOPLE

Abstract

A single small segment (sequence recognition) or domain (conformation recognition) of a protein could act as an antigen (antigenic determinant) vs an antibody. Epitopes of the first kind being a continuous segment along the sequence (linear), generally bent with a typical non-ordered structure (turns and/or loops), can be predicted from the only knowledge of the primary structure. After reviewing the different algorithms, we present PEOPLE (Predictive Estimation Of Protein Linear Epitopes) which uses combined prediction methods, taking into account the basic fundamental properties corresponding to what should be an ideal epitope: bent (secondary structure mainly beta-turns), surface accessible, hydrophilic and mobile and/or flexible. Four classes of basic biophysical parameters are considered for the determination of an antigenic index AG - secondary structure; hydrophilicity; surface accessibility; flexibility. The AG index is finally defined as a linear combination of the four class profiles. Typical applications are presented.