Protein Sequencing with Artificial Intelligence: Machine Learning Integrated Phosphorene Nanoslit

Abstract

AbstractAchieving high throughput protein sequencing at single molecule resolution remains a daunting challenge. Herein, relying on a solid‐state 2D phosphorene nanoslit device, an extraordinary biosensor to rapidly identify the key signatures of all twenty amino acids using an interpretable machine learning (ML) model is reported. The XGBoost regression algorithm allows the determination of the transmission function of all twenty amino acids with high accuracy. The resultant ML and DFT studies reveal that it is possible to identify individual amino acids through transmission and current signals readouts with high sensitivity and selectivity. Moreover, we thoroughly compared our results to those from graphene nanoslit and found that the phosphorene nanoslit device can be an ideal candidate for protein sequencing up to a 20‐fold increase in transmission sensitivity. The present study facilitates high throughput screening of all twenty amino acids and can be further extended to other biomolecules for disease diagnosis and therapeutic decision making.