Quantum Transport Informed Machine Learning Mapping of Current–Voltage Characteristics for Precision Deoxyribonucleic Acid Sequencing

Abstract

Quantum tunneling-based DNA sequencing promises to transform genomic analysis by improving long-read accuracy and enabling high-throughput sequencing, particularly the precise measurement of electrical conductance and tunneling current signatures associated with individual nucleotides. However, key obstacles remain in achieving swift and precise nucleotide identification, such as variation in molecular conductance, noise interference in tunneling current signals, and the complexity of overlapping signal patterns. Here, we employed a quantum transport approach combined with a supervised machine learning (ML) model to accurately classify DNA molecules based on their transmission, conductance, and current readouts, emphasizing their relevance for single-molecule DNA sequencing. This approach significantly resolves overlapping issues with nucleotide classification accuracy as high as 100, 98, and 97% using current, transmission, and conductance readouts, respectively. Sensitivity analysis reveals that current–voltage characteristics are the most effective parameters for distinguishing different nucleotides. Our findings offer a guide for ML mapping of transmission, conductance, and current readouts, enabling rapid and high-precision DNA sequencing. © 2025 Elsevier B.V., All rights reserved.