STRUMP-I: Structure-based machine learning approach to pMHC-I binding prediction using force field energy features.

Area of Science:

• Immunology
• Computational Biology
• Bioinformatics

Background:

• The adaptive immune system uses peptide-MHC class I complexes (pMHC) to detect abnormal cells.
• Accurate prediction of pMHC-I binding is crucial for cancer immunotherapy targeting neoantigens.
• Existing computational methods have limitations in data dependency and generalization across MHC alleles.

Purpose of the Study:

• To develop a novel structure-based computational tool for predicting peptide-MHC class I (pMHC-I) binding.
• To overcome limitations of current sequence-based and structure-based prediction methods.
• To enhance the identification of clinically relevant neoantigen targets for cancer immunotherapy.

Main Methods:

• Developed STRUMP-I (STRUcture-based pMHC Prediction for class I), a novel pMHC binding prediction tool.
• Leveraged a broad set of force-field-derived energy terms as machine-learning features.
• Evaluated performance on independent datasets, including a cancer neoantigen dataset.

Main Results:

• STRUMP-I achieves performance comparable to state-of-the-art sequence-based models.
• STRUMP-I significantly outperforms sequence-based methods on MHC alleles with limited training data.
• Integrating STRUMP-I with sequence-based methods enhances prediction precision.

Conclusions:

• STRUMP-I offers a robust and generalizable approach for pMHC-I binding prediction.
• The tool improves the identification of potential neoantigen targets for cancer immunotherapy.
• STRUMP-I advances the capability to reliably identify clinically relevant neoantigen targets.