Extracting anomalous diffusion parameters from multi-state ensembles of short single molecule trajectories.

Single-molecule tracking measures the stochastic motion of individual biomolecules in the cellular environment. Statistical analysis of trajectory ensembles is required to gain insight into the biophysical nature of mobility states and molecular interactions that they reflect. Mobility states can be parameterized by a generalized diffusion coefficient and anomalous exponent. Experimental constraints such as finite track length and localization precision limit how accurately these parameters can be determined. We compare the performance of analysis methods to recover the input parameters from ensembles of simulated single molecule tracks from different states spanning the range of anomalous diffusive behaviors observed in the cell nucleus. We further develop a framework to quantify error rates in the assignment of mobility states to individual molecules based on recall rates and precision. Our analysis shows that single-track analysis methods are superior to bulk methods in their ability to recover parametric descriptors from mixed populations. The most complete description is obtained by combining outputs from different tools. Our work provides a guide to assess the accuracy of analyses and obtain the most accurate parametric description of experimental single particle tracking data. Statement of significanceExperimental single particle tracking data provides rich insight into molecular interactions directly in living cells. But data analysis depends critically on choosing the correct diffusion model and appropriate tools to extract accurate information. Importantly, it is usually not obvious from the output of a method whether the results are accurate or not. In this work, we use ensembles of tracks simulated with fractional Brownian motion methods to characterize the impact of track length and localization precision on analysis outcomes. We elaborate on specific strengths and weaknesses of commonly used and newly developed analysis tools to provide a template for thorough assessment and quantification of error rates in experimental data analysis.