Machine Learning-Assisted Nanopore for Enhanced Fingerprinting Analysis of Functional Glycans

Precise identification and structural elucidation of glycans have long been a challenge in analytical science. A clear understanding of their structure-function relationship is also crucial for revealing related biological activities. Although emerging nanopore technology has the potential to provide glycan fingerprints at the single-molecule level, direct detection remains significantly difficult due to limitations such as small size, insufficient charge, and efficient methods for nanopore events identification. To overcome these hurdles, we proposed a synergistic enhancement strategy that integrated charge modulation of glycans with the spatial confinement effect of the nanopore lumen. This innovation significantly strengthens the interactions between glycans and the nanopore, thereby amplifying the fingerprint signals of glycans with subtle structural differences. This method not only distinguished between glycan isomers but also enabled qualitative analysis and has been successfully validated in real samples. To further enhance the analytical objectivity, we employed a customized machine learning algorithm to automatically classify and identify translocation events, achieving an overall accuracy of over 97%. In conclusion, this study provides a novel single-molecule sensing perspective for glycan fingerprint analysis.