Deep Learning of Nanopore Sensing Signals Using a Bi-Path Network

Temporal changes in electrical resistance of a nanopore sensor caused by translocating target analytes are recorded as sequence pulses on current traces. Prevalent algorithms for feature extraction pulse-like signals lack objectivity because empirical amplitude thresholds user-defined to single out the from noisy background. Here, we use deep learning based bi-path network (B-Net). After training, B-Net acquires prototypical and ability both pulse recognition without priori assigned parameters. The is evaluated simulated data sets further applied experimental DNA protein translocation. results characterized small relative errors stable trends. shown capable processing with signal-to-noise ratio equal 1, an impossibility threshold-based algorithms. presents generic architecture applicable beyond currents.