Cleveland Clinic and IBM Detail Quantum Machine Learning Model for Cancer Neoantigen Prediction

Framework aims at precision immuno-oncology data bottleneck

Researchers from Cleveland Clinic and IBM Research have developed a quantum machine learning framework designed to predict which tumor gene mutations produce immunogenic neoantigens—abnormal cell-surface proteins capable of triggering a therapeutic T-cell response. Published in Science Advances under the title "Quantum convolutional HLA immunogenic peptide prediction (Q-CHIPP): Next-generation neoantigen prediction with quantum neural network," the joint study combines quantum convolutional neural networks (QCNNs) with biomedical domain knowledge to address the data scarcity and overfitting that have limited classical machine learning in precision immuno-oncology. Predicting immunogenic neoantigens is a central bottleneck for personalized cancer vaccines and targeted immunotherapies because a single tumor can express thousands of mutant candidate peptides, of which only a small fraction are bound by the major histocompatibility complex (MHC) and recognized by T-cell receptors.

Q-CHIPP architecture splits binding and recognition tasks

The Q-CHIPP framework integrates two distinct QCNN models targeting HLA-A*02:01-restricted 9-mer peptides. The first model predicts peptide-MHC binding using anchor-site residues at amino acid positions 2, 4, 5, and 9, while the second predicts T-cell receptor recognition using contact-site residues at positions 3, 5, 6, and 7. To keep binding characteristics from confounding immunogenicity predictions, the TCR model was trained exclusively on experimentally confirmed MHC binders, and a peptide is classified as an immunogenic neoantigen only when both independent QCNN channels return a positive prediction. The architecture was engineered to extract meaningful biological patterns from training sets with as few as 150 samples, directly addressing the data inefficiency of prior approaches.

Execution on pre-fault-tolerant quantum hardware

To run the algorithms on pre-fault-tolerant quantum hardware, the team applied empirical noise mitigation strategies, including Pauli twirling, dynamical decoupling, and controlled shot-based sampling of up to 20,000 shots per circuit execution. A warm-start transfer protocol transferred simulator-optimized weights onto physical quantum processing units for fine-tuning. The researchers scaled the approach to full-length 9-mer peptide modeling across 46 qubits on the IBM Quantum System One deployed at Cleveland Clinic. Under identical parameter constraints and low sample sizes, the error-mitigated QCNN achieved a 6% increase in classification accuracy over parameter-matched classical convolutional neural networks.

Clinical validation on lung cancer cohort

The team validated Q-CHIPP's clinical relevance using a cohort of 111 HLA-A*02:01-positive lung cancer patients treated with immune checkpoint inhibitors, processing 209,889 candidate peptides via a sliding window around somatic mutations. Patient stratification based on Q-CHIPP's predicted immunogenic neoantigen burden showed a statistically significant separation in overall survival (P = 0.0085), outperforming standard NetMHCpan binding rank thresholds (P = 0.0275). The framework identified 25,542 unique candidate immunogenic peptides missed by classical benchmark models and showed strong selection for key TCR-interaction residues at position 7. The open-access study is available in Science Advances, with additional research updates posted on the Cleveland Clinic Lerner Research Institute portal.