Artificially Intelligent Nanogap for Rapid DNA Sequencing: A Machine Learning Aided Quantum Tunneling Approach

Electrical identification of single DNA nucleotides with solid-state nanopore/nanogap and quantum transport technology offers a new paradigm in the field of DNA sequencing and has the potential to supersede existing techniques. However, the overlapping of fingerprint electric conductance signals due to the similar size and comparable frontier orbital energy levels of nucleotides has been a major impediment to identifying them with high accuracy. Herein, a synergistic approach combining the quantum transport method and machine learning algorithms has been devised to achieve the high-precision identification of DNA nucleotides. A model germanene nanogap is investigated for single-nucleotide-based DNA sequencing by calculating the transmission function and current–voltage characteristics. With the transmission function data sets, the Random Forest Classifier algorithm identified all four nucleotides and also demonstrated that binary, ternary, and quaternary combinations of nucleotides could also be classified with a high degree of precision, F1 score, and accuracy. The interelectrode distance analysis illustrates that transmission functions are sensitive to the electrode-nucleotide coupling effect and that the ML classifier can extrapolate the information during classification. Our findings provide a guide to the ML application on the nanogap device to achieve fast, cost-effective, and single-shot nucleotide identification.