Detection of intensity change points in time-resolved single-molecule measurements

We present a method for the analysis of optical single molecule emission data that exhibit discrete intensity jumps. This new method uses a generalized likelihood ratio test that determines the location of an intensity change point based on individual photon arrival times. This test is applied recursively to an entire single molecule intensity trajectory, thus finding each change points. Expectation-maximization clustering and the Bayesian information criterion is then used for accurate determination of the true number of states accessible to the system. This procedure allows rigorous and quantitative determination of intensity change points without the artificial time resolution limitations that arise from binning and thresholding.