What Tools Are Available to Predict Peptide-MHC Binding Affinity?

The Importance of pMHC Binding Affinity

The major histocompatibility complex (MHC) plays a pivotal role in immune defense by recognizing and binding to foreign peptides and presenting them to T cells. MHC class I molecules present peptides to CD8+ cytotoxic T cells, while MHC class II molecules present peptides to CD4+ helper T cells. The specificity and strength of these peptide-MHC (pMHC) interactions determine the efficacy of the resulting immune response.

Read more about the main factors contributing to peptide-MHC binding affinity.

Understanding and predicting pMHC binding affinity and stability is essential in the development of vaccines, cancer immunotherapy, and treatment for autoimmune diseases. Being able to identify peptides that bind strongly to the MHC complex ensures that peptides selected for the development of vaccines or immunotherapies will interact effectively with MHC molecules and elicit the desired immune response.

On the other hand, ensuring that therapeutic proteins do not induce an undesired immune response is critical for drug safety. Bioengineered proteins may contain neoepitopes that could be recognized as foreign by the immune system. Understanding which epitopes could be presented by MHC II molecules, leading to CD4+ T cell activation and the production of anti-drug antibodies, could be important to mitigate potential immunogenicity of new drugs.

Tools to Assess pMHC Binding Affinity

pMHC binding affinity and immunogenicity can be assessed experimentally but given the large variety of antigens and MHC alleles, testing all possible combinations in vitro is often not feasible. Therefore, numerous tools have been developed to facilitate this prediction.

Here’s an overview of the most popular tools for predicting pMHC binding affinity:

Tool

Functionality

Covered species

Allele type

(#of covered alleles)

Input data format

Web interface/download

Ease of use

Good to know

NetMHC 4.0 Binding of peptides to MHC class I molecules

(binding affinity, peptide ranking, binding level)
Human HLA-A
HLA-B
HLA-C
HLA-E
(81)
Plain text format (FASTA or PEPTIDE)
One letter amino acid code
Web interface and download User-friendly interface  Handles large data sets
Works for peptides of any length
Customizable, e.g., specify peptide length, set threshold for strong and weak binders
Mouse
Primate
Pig
Cow
H-2
Patr, Mamu, Gogo
SLA
BoLA
(41)
NetMHCpan 4.1 Pan-specific binding of peptides to MHC class I molecules of known sequence

(binding affinity, peptide ranking, binding level, binding score) 
Human 
HLA-A
HLA-B
HLA-C
HLA-E
Plain text format (FASTA or PEPTIDE)
One letter amino acid code
Web interface and download User-friendly interface  Works for peptides of any length
Customizable, e.g., set threshold for strong and weak binders, include Eluted ligand likelihood
Mouse  H-2
Primates Patr, Mamu, Gogo
Pig  SLA
Cow BoLA
Horse Eqca
NetMHCII 2.3 Binding of peptides to MHC class II molecules
(binding affinity, binding level, binding score) 
Human 
HLA-DR (25)
HLA-DQ (20)
HLA-DP (9)
Plain text format (FASTA)
One letter amino acid code
Web interface and download Straightforward interface Customizable, e.g., set threshold value
Mouse  H-2 (7)
NetMHCIIpan 4.3 Pan-specific binding of peptides to MHC class II molecules of known sequence
(binding affinity, peptide ranking, binding level, binding score) 
Human  HLA-DR
HLA-DQ
HLA-DP
Plain text format (FASTA or PEPTIDE)
One letter amino acid code
Web interface and download Straightforward interface Customizable, e.g., set threshold for strong and weak binders, include Eluted ligand likelihood
Mouse  H-2
Cow BoLA-DRB3
SYFPEITHI Motif-based epitope prediction for peptides binding to MHC class I and II molecules
(method based on binding scores) 
Human  HLA-A
HLA-B
HLA-DR
One letter amino acid code Web interface and download Moderately easy to use Coverage of a variety of MHC class I and II alleles
Estimated epitope prediction reliability for MHC class II approx. 50%
Mouse  H-2
Rat RT1
MHCflurry 2.0 Pan-allele binding of peptides to MHC class I molecules

(binding affinity, peptide ranking, binding score) 
Human and other species (14,000) Command line arguments or CSV file Download Only for trained users, difficult to operate Based on Python
IEDB (immune Epitope Database) Suite of tools for prediction of binding affinities and scores for MHC class I and II, MHC calls I processing and MHC I immunogenicity prediction  Human HLA-A
HLA-B
HLA-C
HLA-E
HLA-G

HLA-DR
HLA-DQ
HLA-DP
Plain text format (FASTA) Web interface and download Comprehensive, straightforward interface, but the large number of options might be overwhelming Customizable: include Eluted ligand likelihood
Mouse H-2
Primate Patr
Mamu
Gogo
Pig  SLA
Cow BoLA
Horse Eqca
Dog DLA

The traditional NetMHC tools (a suite of tools available from DTU, Technical University of Denmark) rely heavily on experimental data, which can be challenging when experimental data is limited for certain alleles. In contrast, the newer NetMHCpan tools are trained on datasets to recognize features that are shared across different alleles, so they are able to generalize across different MHC alleles and not only rely on allele-specific data. These pan tools are often more effective for predicting pMHC binding of novel or rare alleles for which limited experimental data may be available.
At Immudex, we have been using NetMHC for many years to predict pMHC binding affinity before producing custom MHC Monomers or MHC Dextramer® reagents. We evaluate each unique peptide's ability to bind to the requested MHC allele(s) using in silico approaches.

How Do You Validate pMHC Binding?

Whatever tool you choose to predict pMHC binding, you will need to validate the epitope experimentally.

At Immudex we offer a pMHC Binding Screening Service which combines an accurate prediction with experimental validation to ensure reliable identification of strong binders. Find out how this service helps you assess the immunogenicity of your peptides.

Learn more about Epitope Validation.

Future Outlook for pMHC Binding Prediction Tools

The existing prediction tools and algorithms are being continually developed and with the increased use of Artificial Intelligence¹, new and improved pMHC binding prediction tools can be expected over the coming years.

For example, a new algorithm based on machine learning has recently been developed by Ardigen, called ARDisplay-I, which focuses not on pMHC binding, but rather on actual presentation of the peptides. ARDisplay-I predicts the probability of presentation of a given peptide on the cell surface by MHC class I molecules and was able to outperform both NetMHCpan v4.1 and MHCflurry².

References

¹. Borole, P., Rajan, A. Building trust in deep learning-based immune response predictors with interpretable explanations. Commun Biol 7, 279 (2024). https://doi.org/10.1038/s42003-024-05968-2

². Mecklenbräuker et al., Identification of tumor-specific MHC ligands through improved biochemical isolation and incorporation of machine learning. bioRxiv 2024. https://www.biorxiv.org/content/10.1101/2023.06.08.544182v1

Related Resources

Epitope Discovery and Validation

We provide an overview of the main approaches to validate candidate epitopes experimentally.

Factors Contributing to Immunogenicity

What factors play a role in immunogenicity and pMHC complex binding affinity?

MHC Dextramer®

Explore our range of reagents for validating epitopes by pMHC multimer screening.

 

pMHC Binding Screening Service

We offer a service to assess the ability of your candidate peptides to bind to MHC alleles.

Do You Have a Question about pMHC Binding Affinity?

Fill in the form and one of our dedicated specialists will get in touch with you shortly.