Comparing Event Detection Methods in Single-Channel Analysis Using Simulated Data

Abstract

With more states revealed, and more reliable rates inferred, mechanistic schemes for ion channels have increased in complexity over the history of single-channel studies. At the forefront of single-channel studies we are faced with a temporal barrier delimiting the briefest event which can be detected in single-channel data. Despite improvements in single-channel data analysis, the use of existing methods remains sub-optimal. As existing methods in single-channel data analysis are unquantified, optimal conditions for data analysis are unknown. Here we present a modular single-channel data simulator with two engines; a Hidden Markov Model (HMM) engine, and a sampling engine. The simulator is a tool which provides the necessary a priori information to be able to quantify and compare existing methods in order to optimize analytic conditions. We demonstrate the utility of our simulator by providing a preliminary comparison of two event detection methods in single-channel data analysis; Threshold Crossing and Segmental k-means with Hidden Markov Modelling (SKM-HMM).