The automated analysis of data from single ionic channels

Abstract

The development of single channel recording has brought with it the need to analyse enormous amounts of data. The data analysis is time consuming and subject to observer biases since the events are random in time and are contaminated with uncorrelated noise. We have developed a heuristic pattern recognition program which identifies with high precision single channel currents and rejects contaminating noise. The program interactively provides for a variety of amplitude and duration measures. Analysis is flexible and rapid: a file containing over 10,000 events can be analysed in under 2 h. Specific detection features include variable lowpass filtering, automatic baseline restoration, and adaptive amplitude thresholds. A record is analysed through duration histograms, binomial estimates of the number of active channels present, cross-correlation estimates between parameters, spectral analysis of events and background noise, and stationarity of mean channel current. The graphic output facilities can plot raw data (after filtering and baseline restoration) with the idealized signal superimposed or with detected events underlined. A batch processing facility has been included to allow processing of data during periods of low computer demand.