IMPROVE: a feature model to predict neoepitope immunogenicity through broad-scale validation of T-cell recognition

Area of Science:

• Immunology
• Computational Biology
• Oncology

Background:

• Mutation-derived neoantigens are key targets for cancer immunotherapy.
• Improved neoepitope identification and prediction tools are needed for effective neoepitope targeting.
• Current computational tools face limitations in predicting T cell recognition due to data scarcity.

Purpose of the Study:

• To develop a predictive model for neoepitope immunogenicity.
• To enhance the identification of patient-specific neoantigens for cancer immunotherapy.

Main Methods:

• Utilized a dataset of 17,500 experimentally validated neoepitope candidates from 70 cancer patients.
• Evaluated 27 distinct neoepitope characteristics.
• Developed a random forest model named IMPROVE for predicting neoepitope immunogenicity.

Main Results:

• The IMPROVE model significantly advanced neoepitope identification compared to existing methods.
• Hydrophobic and aromatic residues within the peptide binding core were identified as critical features for predicting immunogenicity.
• The model demonstrated improved capacity to predict T cell recognition of neoepitopes.

Conclusions:

• The IMPROVE model offers a significant advancement in predicting neoepitope immunogenicity.
• Accurate prediction of neoepitopes enhances the potential for successful cancer immunotherapy strategies.
• This work provides a valuable tool for selecting potent neoantigens for therapeutic development.