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

Area of Science:

• Analytical Chemistry
• Biotechnology
• Molecular Biology

Background:

• Glycan identification and structural elucidation are critical for understanding biological functions.
• Nanopore technology offers potential for single-molecule glycan analysis but faces challenges like small size and low charge.
• Existing methods struggle with distinguishing subtle structural differences in glycans.

Purpose of the Study:

• To develop a synergistic strategy for enhanced single-molecule glycan detection and analysis.
• To overcome limitations of nanopore technology for glycan fingerprinting.
• To improve the accuracy and objectivity of glycan identification.

Main Methods:

• Integrated charge modulation of glycans with nanopore spatial confinement.
• Utilized nanopore events identification for signal amplification.
• Employed a customized machine learning algorithm for automated event classification.

Main Results:

• Successfully distinguished between glycan isomers with subtle structural differences.
• Achieved signal amplification, enhancing glycan fingerprint sensitivity.
• Validated the method in real samples with over 97% accuracy using machine learning.

Conclusions:

• The developed method offers a novel single-molecule sensing approach for glycan fingerprint analysis.
• Synergistic enhancement strategy significantly improves glycan detection and structural elucidation.
• Machine learning integration enhances analytical objectivity and identification accuracy.