Decoding Both DNA and Methylated DNA Using a MXene-Based Nanochannel Device: Supervised Machine-Learning-Assisted Exploration
Sneha Mittal, Souvik Manna, Milan Kumar Jena, Biswarup Pathak
Abstract
An important pursuit in medical research is to develop a fast and low-cost technique capable of sequencing the entire human genome (DNA) and epigenome (methylated DNA) de novo. Such a method would enable the advancement of personalized medicines and a universal cancer screening test. In this regard, we introduce a novel supervised machine learning (ML) approach for ultrarapid prediction of transmission function of DNA and methylated DNA nucleobases using a MXene-based nanochannel device. The proposed device can detect the targeted nucleobases with good transmission sensitivity. The random forest regression (RFR) model can predict the transmission function of each unknown nucleobase with root-mean-square error (RMSE) values as low as 0.16. Interestingly, if the machine is trained with the dataset of methylated DNA nucleobases, it can selectively identify all four DNA nucleobases with good accuracy. Therefore, our study demonstrates an effective approach for quick and accurate whole-genome and epigenome sequencing applications.