Major histocompatibility complex(MHC, also named as HLA in human being, short for Human leukocyte antigen) molecules play a very important role in the adaptive immune system. MHC molecules bind short peptides from antigens and present them on the surface of a cell for recognition by T-cell receptors (TCR). The presented peptide and MHC complexes induce the naive T cells to proliferate and differentiate into armed effector T cells that help to remove the antigens. The discovery of which peptides binding to a given MHC allele is very helpful for us in understanding the mechanism of adaptive immune response, as well as developing peptide based vaccine. The experimental identification of peptide binding affinity is a time-consuming and expensive process. In contrast, computational approaches for MHC peptide binding prediction can help immunologists to select a small set of promising epitope candidates for further investigations. A precise prediction method can thus reduce the cost significantly.
     MHC molecules can be divided into two subclasses, MHC class I (for human being, mainly HLA-A, HLA-B, and HLA-C) and MHC class II (for human being, mainly HLA-DP, HLA-DQ, and HLA-DR). MHC class II molecules present peptides from exogenous resources to CD4 helper T cell. MHC molecules are highly polymorphic. For example, there are hundreds of variants in HLA-DR alleles. However, only around 20 of them have several hundred or more experimental verified binders and non-binders. How to predict the binding specificity of MHC molecules with very few or no training data thus becomes a very challenging problem. Currently three computational methods, TEPITOPE [1], NetMHCIIpan [2, 3] and MultiRTA [4], have been developed to make MHC II pan-specific predictions. Compared with the other two methods, TEPITOPE uses PSSM (Position-specific scoring matrix) to score the peptides, which can be easily understood and explained. It has been reported to achieve decent performance in identifying MHC class II ligands, T-cell epitopes, and peptide binding cores. However, TEPITOPE can only make predictions for 50 HLA DRB alleles using 50 PSSMs built from 35 experimentally obtained profiles. By measuring the pocket similarity of different MHC molecules [5], we develop TEPITOPEpan to extend the original TEPITOPE method to peptide binding prediction to more alleles. Generally speaking, it can make the qualitative peptide binding prediction to any HLA-DR molecule of known sequence.

Matrices DownLoad

    This file contains over 700 (734) TEPITOPEpan scoring matrices derived from PSSMs in TEPETOPE. To download the matrices in zip format:


Epan-Set4 Dataset

    The evaluation data set contains 2412 binary data points (1256 binders, threshold=500nM) covering 14 HLA-DR alleles, retrieved from IEDB database of Mar., 2011. To download the dataset, click this button here:



  1. Sturniolo T, Bono E, Ding J, Raddrizzani L, Tuereci O, Sahin U, Braxenthaler M, Gallazzi F, Protti M P, Sinigaglia F, Hammer J. 1999. Generation of tissue-specific and promiscuous HLA ligand database using DNA microarrays and virtual HLA class II matrices. Nat. Biotechnol. 17:555-561.
  2. Nielsen M, Justesen S, Lund O, Lundegaard C, Buus S. 2010. NetMHCIIpan-2.0 - Improved pan-specific HLA-DR predictions using a novel concurrent alignment and weight optimization training procedure. IMMUNOME RESEARCH . 6:9.
  3. Nielsen M, Lundegaard C, Blicher T, Peters B, Sette A, Justesen S, Buus S, Lund O. 2008. Quantitative Predictions of Peptide Binding to Any HLA-DR Molecule of Known Sequence: NetMHCIIpan. PLoS computational biology. 4(7):e1000107.
  4. Bordner AJ, Mittelmann HD. 2010. MultiRTA: a simple yet reliable method for predicting peptide binding affinities for multiple class II MHC allotypes. BMC Bioinformatics. 11:482.
  5. Zhang H, Lund O, Nielsen M. 2009. The PickPocket method for predicting binding specificities for receptors based on receptor pocket similarities: application to MHC-peptide binding. Bioinformatics. 25:1293-1299.