The MHC class II binding prediction tools can be found at http://tools.immuneepitope.org/main/html/tcell_tools.html.
A RESTful interface is also available for MHC class I and class II prediction tools. This allows users to perform predictions on the IEDB server in batch mode without having to install any software on their own systems. Additionally, users are always assured that they are using the latest version of the tools.
Peptide Binding to MHC Class II Molecules
Users can select from eight different methods for predicting class II epitopes – IEDB recommended, ARB, SMM-align, Sturniolo, Combinatorial Library, Consensus, NN-align, and NetMHCIIpan. By default, the overall best method (IEDB recommended) is selected. However, not all methods can currently make predictions for all alleles, so only the alleles available will be displayed. The eight methods are described further below.
Average Relative Binding (ARB)
Average relative binding (ARB) matrix binding prediction method is based on the assumption that each residue along the peptide molecule independently contributes to binding affinity. When a residue R occurs at position i in the peptide, it is assumed to contribute a constant amount of Ri to the free energy of binding of the peptide. The effect of each of the 20 possible amino acids at each possible position along the peptide sequence, therefore, can be estimated by a matrix of coefficients. The overall binding propensity of each peptide sequence, an algorithm “score”, is calculated by multiplying the Ri coefficients. Predicted IC50 values, which provide quantitative KD (IC50) predictions, are then calculated by mathematical transformations of the algorithm scores (Bui et al., Immunogenetics 2005).
The MHC class II binding groove is open at both ends making the correct alignment of a peptide in the binding groove a crucial part of identifying the core of an MHC class II binding motif. The stabilization matrix alignment method, SMM-align, allows for direct prediction of peptide:MHC binding affinities. The method uses amino terminal peptide flanking residues (PFR) to get a consistent gain in predictive performance by favoring binding registers with a minimum PFR length of two amino acids. The method predicts quantitative peptide:MHC binding affinity values. The method has been trained and evaluated on a data set that covers the nine HLA-DR supertypes suggested and three mouse H2-IA allele. The method is described by Nielsen et al. (BMC Bioinformatics, 2007).
This matrix-based approach is used in the TEPITOPE class II epitope prediction program. It is described in Sturniolo et al. (Nat. Biotechnol., 1999).
The positional scanning combinatorial libraries approach utilized a pool of random peptide libraries to systematically measure the contribution to MHC binding from each amino acid at each of the nine positions at the binding peptide. Each pool in the library contains 9-mer peptides with one fixed residue at a single position. With each of the 20 naturally occurring residues represented at each position along the 9-mer backbone, the entire library consisted of 180 peptide mixtures. Competitive binding assays were then carried out to determine the IC50 values for each pool. IC50 values for each mixture were standardized as a ratio to the geometric mean IC50 value of the entire set of 180 mixtures, and then normalized at each position so that the value associated with the optimal value at each position corresponds to 1. For each position, an average (geometric) relative binding affinity (ARB) was calculated, and then the ratio of the ARB for the entire library to the ARB for each position was derived. The final results are a set of 9 by 20 scoring matrices which could predict the binding of novel peptides to MHC molecules.
The methodology is the same as that used for the MHC class I combinatorial library tool as described in Sidney et al. Immunome Res. 2008. A paper specifically describing the class II tool will be published in 2010.
The consensus method was developed by the IEDB team by exploiting features of the other three aforementioned methods. The method was updated with the introduction of NN-align, so the revised Consesnus method uses NN-align, SMM-align, and the combinatorial peptide scanning library. When the scanning library is not available for an allele, the Sturniolo method is used instead. A paper describing the original method was published by Wang et al. (PLoS Comput Biol, 2008).
NN-align is an artificial neural network-based alignment algorithm for MHC class II peptide binding prediction. It simultaneously identifies the MHC class II binding core and binding affinity. The method is trained using an algorithm that corrects bias in the training data caused by redundancy in binding core representation. Prediction accuracy has been shown to improve significantly when information about the residues flanking the peptide-binding core is taken into account. A 2009 paper in BMC Bioinformatics by Nielsen and Lund describe the method in detail (PMID: 19765293).
NetMHCIIpan predicts binding of peptides for over 500 HLA-DR alleles using artificial neural networks. A paper describing the method was published by Nielsen et al. in Immunome Research, November 2010.
The datasets used in assessing the performance of the ARB, SMM-align, and Sturniolo methods and in developing the Consensus method, as described in Wang et al., can be found at http://mhcbindingpredictions.immuneepitope.org/MHCII/. The three datasets can be used for developing algorithms that predict peptides binding to MHC class II molecules and/or activating CD4+ T cells. The first is a comprehensive dataset consisting of more than 10,000 previously unpublished MHC-peptide binding affinities for 16 alleles (peptide_affinity_dataset.zip). The second dataset is a text file of 29 peptide/MHC crystal structures found in the PDB that can be used for binding core predictions (non_redundant_pdb_core_pep_allele.txt). The third dataset contains 664 peptide sequences experimentally tested for CD4+ T-cell responses (LCMV_T_cell_activation.txt).
IEDB recommended is the default prediction method selection. It considers all alleles and their corresponding peptide lengths for a particular species. For each allele-length combination, consensus method is used, which includes NN, SMM, and CombLib. If none of these methods are available for the allele, NetMHCIIpan is used.