Machine Learning Boosted Quantum-Profiling of Blood Antigens

The molecular-level characterization of glycans, a challenging yet highly desirable goal, is crucial for the comprehensive advancement of glycosciences. Despite significant advances in analytical techniques, including NMR and mass spectrometry, structural and configurational complexity hinders the ability to identify carbohydrates, especially high-order saccharides with regioisomeric glycosidic linkages. In this article, we present a computational methodology that utilizes a quantum tunneling method coupled with machine learning (ML) to recognize a wide range of blood antigens simultaneously. Random forest classifier with SHapley Additive exPlanations (SHAP) interpretability performs rapid quantum profiling of all considered molecules with good precision and sensitivity. Our proposed ML-enhanced quantum methodology offers a powerful alternative to conventional techniques, facilitating accurate and high-throughput characterization of carbohydrates by performing "sugar calling" from their transmission signatures.