Optimize prediction sensitivity by following our recommendations:
Based on a dataset comprised of >750 experimentally validated MHC-peptide binding affinities for the proteins HPV16 E6 and HPV16 E7, we evaluated the performance of all prediction servers included in MHCcombine.
The detailed results of our analysis can be accessed via:
M. Bonsack, S. Hoppe, J. Winter, D. Tichy, C. Zeller, M. D. Küpper, E. C. Schitter, R. Blatnik, A. B. Riemer
Performance Evaluation of MHC Class-I Binding Prediction Tools Based on an Experimentally Validated MHC-Peptide Binding Data Set.
Please note: NetMHCpan 4.1 was published after our analysis and is therefore not contained in our recommendations table.
We would recommend to use the following servers for the following purposes:
pooled (8-11-mers)
8-mers
9-mers
10-mers
11-mers
HLA-A1
NetMHCpan 4.0
NetMHC 3.4
NetMHCpan 4.0
NetMHC 3.4
NetMHCpan 4.0
HLA-A2
NetMHCpan 4.0
NetMHC 4.0
NetMHCpan 4.0
IEDB SMMPMBEC
NetMHC 4.0
HLA-A3
NetMHC 3.4
NetMHC 4.0
NetMHC 3.4
IEDB recommended
IEDB SMMPMBEC
HLA-A11
IEDB recommended
NetMHCpan 4.0
NetMHC 4.0
IEDB recommended
IEDB recommended
HLA-A24
NetMHCcons 1.1
NetMHCpan 4.0
NetMHCpan 4.0
NetMHCcons 1.1
IEDB recommended
HLA-B7
NetMHCcons 1.1
IEDB recommended
IEDB recommended
NetMHCpan 3.0
NetMHC 3.4
HLA-B15
NetMHCpan 4.0
NetMHCpan 3.0
NetMHCpan 3.0
NetMHC 4.0
NetMHCpan 4.0
We found that many peptides are missed applying the thresholds recommended by the servers. To increase prediction sensitivity we recommend using the following more tolerant thresholds. Click on the prediction method to expand a table of thresholds.