Inferring chromatin looping interactions from microscopy-based measurements: opportunities and pitfalls

Chromatin looping plays essential roles in several nuclear processes and live-cell microscopy has become a key method for studying how chromatin loops behave in time. Extracting dynamical parameters including looping probability, lifetime, and search time from live-imaging data requires inferring looping events from noisy 3D distance trajectories. By analyzing polymer simulation data with known ground truth, we show that with realistic localization precision, the common method of simple thresholding on 3D distance measurements can result in severe misestimation of looping probability, lifetime, and search time. We use simple statistical considerations to show that inference from simple thresholding becomes infeasible when localization error is comparable to the size of the interaction to be studied. Finally, we demonstrate how noise-aware methods can be used to estimate looping parameters more accurately in these regimes.