Assessing Neuron Response to External Stimuli with a Data-Driven Procedure for Spike Train Extraction and GAMLSS Regressions

Gaining a better understanding of the brain is an important aspect in biological research. To this end, new measurement tools are employed to collect data on calcium traces for mice to analyze the neuron response to external stimuli. Suitable statistical analysis methods for the challenging data sets emerging from these measurements are essential to eventually learn more about the functioning of the brain. We use the calcium traces collected for one mouse across three sessions to showcase a statistical analysis focusing on extracting potential effects of external stimuli on the neuron response based in particular on advanced regression modeling approaches. This statistical analysis consists of: (1) a heuristic algorithm for spike train extraction, (2) an exploratory analysis in combination with suitable pre-processing to highlight features of the data and (3) a regression analysis based on zero-adjusted generalized additive linear models for location, scale and shape to relate the spike trains to stimuli and sessions as well as inspect their evolvement over time. Results indicate how the flexibility of advanced regression modeling approaches allows to account for specific data structures as well as easily compare different modeling approaches.