2 COMPLEMENTARY METHODS FOR PREDICTING PEPTIDES BINDING MAJOR HISTOCOMPATIBILITY COMPLEX-MOLECULES

Peptides that bind to major histocompatibility complex products (MHC) are known to exhibit certain sequence motifs which, though common, are neither necessary nor sufficient for binding: MHCs bind certain pepti des that do not have the characteristic motifs and only about 30% of t he peptides having the required motif, bind. In order to develop and t est more accurate methods we measured the binding affinity of 463 nona mer peptides to HLA-A2.1. We describe two methods for predicting wheth er a given peptide will bind to an MHC and apply them to these peptide s. One method is based on simulating a neural network and another, cal led the polynomial method, is based on statistical parameter estimatio n assuming independent binding of the side-chains of residues. We comp are these methods with each other and with standard motif-based method s. The two methods are complementary, and both are superior to sequenc e motifs. The neural net is superior to simple motif searches in elimi nating false positives. Its behavior can be coarsely tuned to the stre ngth of binding desired and it is extendable in a straightforward fash ion to other alleles. The polynomial method, on the other hand, has hi gh sensitivity and is a superior method for eliminating false negative s. We discuss the validity of the independent binding assumption in su ch predictions. (C) 1997 Academic Press Limited.