Evaluation of MHC-II Peptide Binding Prediction Servers for Vaccine Research
Author Information
Author(s): Lin Hong Huang, Zhang Guang Lan, Tongchusak Songsak, Reinherz Ellis L, Brusic Vladimir
Primary Institution: Dana-Farber Cancer Institute
Hypothesis
How accurately can different prediction servers identify MHC-II peptide binding?
Conclusion
Current HLA-II prediction servers have limited accuracy compared to HLA-I predictors, indicating a need for improved methods.
Supporting Evidence
- The study assessed a total of 113 predictors across various HLA-DR alleles.
- 17 predictors showed good performance, while 55 showed poor performance.
- The best individual predictor was NETMHCIIPAN, followed by PROPRED and IEDB (Consensus).
- Current predictive capabilities allow for the prediction of only 50% of actual T-cell epitopes.
Takeaway
This study looked at different computer programs that predict how well certain proteins can bind to immune system molecules. It found that many of these programs don't work very well.
Methodology
The study analyzed the performance of 21 HLA-II binding prediction servers using data from 721 peptide binding assays.
Potential Biases
The performance assessments may be biased due to the use of pre-defined peptide sets rather than standardized full-overlapping studies.
Limitations
The study was limited to seven common HLA-DR molecules and did not assess all possible predictors.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website