Compactness regularization in the analysis of dipolar EPR spectroscopy data.

Dipolar electron paramagnetic resonance (EPR) experiments, such as double electron-electron resonance (DEER), measure distributions of nanometer-scale distances between paramagnetic centers, which are valuable for structural characterization of proteins and other macromolecular systems. One challenge in the least-squares fitting analysis of dipolar EPR data is the separation of the inter-molecular contribution (background) and the intra-molecular contribution. For noisy experimental traces of insufficient length, this separation is not unique, leading to identifiability problems for the background model parameters and the long-distance region of the intra-molecular distance distribution. Here, we introduce a regularization approach that mitigates this by including an additional penalty term in the objective function that is proportional to the variance of the distance distribution and thereby penalizes non-compact distributions. We examine the reliability of this approach statistically on a large set of synthetic data and illustrate it with an experimental example. The results show that the introduction of compactness can improve identifiability.