Optimized threshold selection for single-molecule two-color fluorescence coincidence spectroscopy.

Two-color coincidence detection is a single-molecule fluorescence technique that is capable of resolving subpopulations of biomolecular complexes at very low concentrations. In this paper, we have developed a method that automatically determines appropriate thresholds for the analysis of sets of two-color coincidence data. This has the distinct advantage of allowing the rapid determination of optimized thresholds in a reproducible fashion. The trade-offs involved in such selections are that the thresholds should be high enough both to exceed the background photon count rates and to ensure a low rate of chance coincident events and that they should be low enough to give reasonably high rates of fluorescence events. Previously, thresholds were selected by judgment to balance these various separate considerations. The method reported in this paper incorporates the three factors into the maximization of a single value derived from the data as a function of the thresholds used in the two channels. The value that is maximized is a ratio of event rates, specifically the rate of coincident events above that expected by chance, divided by the total event rate; this is called the association quotient. In this paper, we demonstrate that maximization of the association quotient selects appropriate thresholds for data derived from dual-labeled duplex DNA samples over a range of concentrations and laser powers. This method should allow the application of two-color coincidence detection to more complex biological systems and cells where the sample concentration and background levels are more variable and where it is not possible to run separate control experiments to determine the latter statistics.