Speaker
Description
Nanopore-based resistive pulse sensing is an experimental technique that allows characterizing proteins on a single-molecule level. The signal from the nanopore depends on the characteristics of the pore, as well as the shape, volume, and orientation of the protein.
To investigate those current traces, we employed a machine learning algorithm trained on bead-model-based simulation data. Our results revealed that the machine learning approach can extract more information from a single translocation trace compared to the conventional approach (extracting protein shape with an accuracy of 80%). This investigation points toward the most important features of the signal that enables the characterization of a protein in nanopore-based resistive pulse sensing.
