Théo Gauvrit is a graduate student at the University of Bordeaux, in the lab of Andreas Frick at the Neurocentre Magendie in Bordeaux, France. He first joined the team in February of 2020 for his master thesis in bioinformatics, during which he developed tools for the analysis of electrophysiological data to characterize atypical sensory information processing in the Fmr1KO mouse model of autism. He continued his work in the lab as a doctoral candidate, studying the role of neuronal noise and network states in the generation of sensory response variability. He is interested in better understanding the role of neuronal noise in atypical sensory processing in autism and exploring this feature as a translational biomarker of autism and a targetable mechanism for drug development. Based on his results from the analysis of single neuron recordings, he is now aiming to investigate the impact of noise and neuronal variability on neocortical processing at the neuronal population level. To this end, he is closely collaborating with experimental neuroscientists to analyze neuronal population activity from calcium imaging and multielectrode recordings. He is also aiming to obtain a model of circuit function using machine learning, in order to tackle the different dynamic components of neuronal activity involved in atypical sensory information processing in different mouse models of autism.

Principal Investigator: Andreas Frick

Fellow: Adrien Corniere

Undergraduate Fellow Project: Machine learning tools to explore atypical sensory information processing in autism mouse models

Sensory alterations were recently added as one of the core features of autism. These sensory characteristics can be exploited to better understand the neurobiological mechanisms of autism and to decipher atypical neuronal computing in autism. Preclinical models of autism allow us to study neuronal activity underlying sensory processing by employing state-of-the-art techniques like calcium imaging and combining them with behavioral tasks.
This project aims to explore atypical sensory perception in autism, focusing on neuronal population dynamics. We will employ machine learning tools to analyze complex multidimensional data obtained by combining calcium imaging in the somatosensory cortex with a perceptual decision-making task. The development of different pipelines in Python will allow the detailed exploration of neuronal activity during two fundamental phases of perception: the encoding of the stimulus and the decision-making (response) of the mouse. In addition, these results will permit us to test our previous hypothesis on increased neuronal noise and network imbalance in autism. This project will be the first step to the construction of a Generalized Linear Models (GLM) that will allow us to tackle the different dynamic components of neuronal activity involved in atypical sensory processing in different mouse models of autism.