Normally functioning cells process and continually present on their surface, short peptides bound to class I major histocompatibility complex (MHC) molecules. When the peptides are products of a foreign genome, such as the genome of a virus that has infected the cell, the complex is recognized by cytotoxic T cells, and a series of events is triggered, which terminates in cell death. The ability to determine which protein segments will bind particular MHC molecules with high  affinity is therefore of importance for the development of peptide vaccines. 

We introduce a new method based on linear programming for predicting HLA-A2 binding peptides.  The method has a sensitivity ~ 0.75, and specificity ~ 0.90 based upon a jack-knife procedure. The major approximation is that contribution of each position to the overall binding free energy depends only on the amino acid at that position. The database employed to develop the method uses the binding affinities for 536 9-residue-long peptides and 457 10-residue-long peptides. 

This server is an implementation of the method.  It can perform the following three tasks: 

1. Predict the ln(IC50) of a peptide based on its sequence. 
2. prediction of the best HLA-A2 binding peptide(s) from an input protein sequence.
3. Prediction of mutant peptide(s) that bind better than the wild type while sharing its T cell epitope.

Peters, B., Tong, W., Sidney, J., Sette, A. & Weng, Z. (2003)
Examining the Independent Binding Assumption for Binding of Peptide Epitopes to MHC-I Molecules
Bioinformatics. 19:1765-1772.
Abstract PDF

Supplemental Materials

SMM Pair Coefficents

Contact:

Zhiping Weng, zhiping.weng@umassmed.edu 

Funding Support:

This work was partially supported by Whitaker research grant RG-00-0426 and NSF grant 0078194 awarded to Zhping Weng.