Читать книгу Informatics and Machine Learning. From Martingales to Metaheuristics онлайн

38 страница из 101

The Nanoscope described in ssss1 builds from nanopore detection with introduction of reporter molecules to arrive at a nanopore transduction detection paradigm. By engineering reporter molecules that produce stationary statistics (a SCW) together with ML signal analysis methods designed for rapid analysis of such signals, we arrive at a functioning “nanoscope.”

Nanopore detection is made possible by the following well‐established capabilities: (i) classic electrochemistry; (ii) pore‐forming protein toxin in a bilayer; and (iii) patch clamp amplifier. Nanopore transduction detection leverages the above detection platform with (iv) an event‐transducer pore‐blockader that has stationary statistics and (v) ML tools for real‐time SCW signal analysis. The meaning of “real‐time” is dependent on the application. In the Nanoscope implementation discussed in ssss1, each signal is usually identified in less than 100 ms, where calling accuracy is 99.9% if rejection is employed, and improved even further if signal sample duration, when a call is forced, is used with duration greater than 100 ms.