Data-driven investigation of the heterogeneity of neuronal biophysical properties in the cerebral cortex

Heterogeneity of neuronal biophysical properties is thought to play a crucial role in network properties. Yet a precise qualification and quantification of these neuronal properties' heterogeneity is still lacking. To leverage from the increasing number of publicly available database of intracellular electrophysiological recordings, we designed a database-structure independent data treatment pipeline, with the aim of overcoming the difference in databases organization, compensating for difference in curation, as well as proposing characterization methods that are versatile and robust enough to account for the intrinsic diversity of neuronal properties. We used this pipeline to gather multiple databases of electrophysiological recordings and characterize the heterogeneity of biophysical properties according to different factors (e.g.: cellular type, cortical layer, cortical area, animal...), while limiting the variance due to difference originating from database-specific acquisition protocols. This cross database analysis of neuronal diversity represents a crucial step toward the establishment of biophysically constrained neuronal network models. Indeed, heterogeneity is known to represent a crucial aspect of neuronal network, yet it has mainly been studied using theoretical variance (i.e.: increasing the variance around the mean for particular properties). Therefore, precisely reproducing electrophysiological heterogeneity at the neuronal level in a network model, can provide valuable insights as to the multiscale interaction mechanisms supporting the emergence of more complex network dynamical properties. The overall approach in the work is guided in large part by Open Science principles, thus the conceptual goal of exploiting results at large from the neuroscience community, and the practical goal of providing easily applicable software tools to that community. As such the thesis discusses the evolving state of Open Science, highlighting the diverse efforts to establish common tools and frameworks to facilitate better collaboration among neuroscientists.