Accurate modeling of peptide-MHC structures with AlphaFold

Area of Science:

• Structural immunology
• Computational biology
• Molecular modeling

Background:

• Major histocompatibility complex (MHC) proteins are crucial for T cell surveillance by presenting peptides on cell surfaces.
• Accurate in silico prediction of peptide-MHC interactions is vital for structural immunology and therapeutic development.
• Current prediction tools face limitations in modeling accuracy for both class I and class II MHC molecules.

Purpose of the Study:

• To develop and validate a novel AlphaFold-based computational pipeline for predicting the 3D structures of peptide-MHC complexes.
• To assess the accuracy of the developed pipeline for both class I and class II MHC molecules.
• To demonstrate the utility of the pipeline in analyzing cancer neoantigens and improving peptide-MHC binding predictions.

Main Methods:

• Utilized an AlphaFold-based pipeline to predict the three-dimensional structures of peptide-MHC complexes.
• Applied the pipeline to both class I and class II MHC molecules.
• Validated the prediction accuracy against existing computational tools and experimental data, including a cancer neoantigen/wild-type peptide pair.

Main Results:

• The AlphaFold-based pipeline achieved high accuracy in predicting peptide-MHC complex structures.
• The method outperformed existing tools in modeling accuracy for class I MHC and peptide register prediction for class II MHC.
• Experimental validation confirmed the pipeline's performance and utility, particularly for cancer-related peptide-MHC interactions.

Conclusions:

• The developed AlphaFold-based pipeline offers a significant advancement in the in silico prediction of peptide-MHC complex structures.
• This tool enhances the ability to model these critical molecular interactions, aiding structural immunology research.
• The pipeline shows promise for applications in areas such as cancer neoantigen identification and optimizing peptide-MHC binding predictions for drug discovery.