PREVENTING ERRORS WHEN ESTIMATING SINGLE-CHANNEL PROPERTIES FROM THE ANALYSIS OF CURRENT FLUCTUATIONS

The conductance, number, and mean open time of ion channels can be est imated from fluctuations in membrane current. To examine potential err ors associated with fluctuation analysis, we simulated ensemble curren ts and estimated single channel properties. The number (N) and amplitu de (i) of the underlying single channels were estimated using nonstati onary fluctuation analysis, while mean open time was estimated using c ovariance and spectral analysis. Both excessive filtering and the anal ysis of segments of current that were too brief led to underestimates of i and overestimates of N. Setting the low-pass cut-off frequency of the filter to greater than five times the inverse of the effective me an channel open time (burst duration) and analyzing segments of curren t that were at least 80 times the effective mean channel open time red uced the errors to <2%. With excessive filtering, Butterworth filterin g gave up to 10% less error in estimating i and N than Bessel filterin g. Estimates of mean open time obtained from the time constant of deca y of the covariance, tau(obs), at low open probabilities (P(o)) were m uch less sensitive to filtering than estimates of i and N. Extrapolati ng plots of tau(obs) versus mean current to the ordinate provided a me thod to estimate mean open time from data obtained at higher P(o), whe re tau(obs) no longer represents mean open time. Bessel filtering gave the least error when estimating tau(obs) from the decay of the covari ance function, and Butterworth filtering gave the least error when est imating tau(obs) from spectral density functions.