Deep learning for nanopore ionic current blockades

DNA molecules can electrophoretically be driven through a nanoscale opening in a material, giving rise to rich and measurable ionic current blockades. In this work, we train machine learning models on experimental ionic blockade data from DNA nucleotide translocation through 2D pores of different diameters. The aim of the resulting classification is to enhance the read-out efficiency of the nucleotide identity providing pathways toward error-free sequencing. We propose a novel method that at the same time reduces the current traces to a few physical descriptors and trains low-complexity models, thus reducing the dimensionality of the data. We describe each translocation event by four features including the height of the ionic current blockade. Training on these lower dimensional data and utilizing deep neural networks and convolutional neural networks, we can reach a high accuracy of up to 94% in average. Compared to more complex baseline models trained on the full ionic current traces, our model outperforms. Our findings clearly reveal that the use of the ionic blockade height as a feature together with a proper combination of neural networks, feature extraction, and representation provides a strong enhancement in the detection. Our work points to a possible step toward guiding the experiments to the number of events necessary for sequencing an unknown biopolymer in view of improving the biosensitivity of novel nanopore sequencers.

Automated summary

In this study, machine learning models were trained on experimental data to improve the accuracy of DNA nucleotide sequencing. By reducing the data dimensionality and utilizing deep neural networks, a novel method was proposed to enhance the read-out efficiency, reaching high accuracy levels.