Prediction of Neuronal Firing Patterns in Zebrafish Embryos Using PCA

Abstract

Neurobiology has long been concerned with identifying patterns in neuronal activity in order to elucidate mechanisms of brain function and reaction. Novel technologies have recently enabled analysis of whole neuronal networks at the single-cell level for organisms as complex as zebrafish. An available dataset utilizing these new imaging techniques was published and analyzed by Chen et al, who exposed zebrafish to a variety of stimuli and analyzed neuronal responses with light-sheet imaging. These data were split into stimulus and whole-fish datasets, from which principal components were generated using Scikit-Iearn. Principal components analysis (PCA) were compared to one another based on their locations and time-series, and no correlation between the similarity of locations and time-series for each component pair was found despite generation of expected results for each metric individually. This is a novel finding that implies that there are significant patterns in neuronal activity that are not dependent on localization within a fish.