From vision to action: Systems analysis of sensorimotor circuitry controlling visually-guided behaviour (360G-Wellcome-101195_Z_13_A)

Despite the emergence of technologies to record from thousands of neurons during animal behaviour, two major issues are currently limiting the transfer and utilisation of ultra-high-dimensional datasets between neuroscience groups. First, raw data files are often enormous, making data transfer inconvenient and expensive. Second, there is no common format for storing behavioural and neurophysiological data, making it challenging to interpret and work with data from another lab. By the end of this project, we will have provided the neuroscience community with a platform that can solve both challenges. Based on our experience, along with input from other neuroscience labs at UCL, we will develop a universal “SysNeuro” format for pre-processed, multi-modal, neurophysiology and behavioural data. This will enable labs to exchange compact files that contain all data and metadata relevant to the experiment along with full details of pre-processing algorithms. We will provide open-source software to easily build and manage SysNeuro files and execute advanced data analyses, including deep-learning to discover network dynamics that predict behaviour. Through a symposium at UCL, we will promote the uptake of our platform, provide hands-on training and create a forum to discuss current and future priorities related to data analysis and sharing. Overall, this project will (1) improve utilisation of light-sheet imaging data from our main project, (2) enhance data sharing across the systems neuroscience community and (3) provide easy-to-use tools that lower the barrier to entry for other researchers to flexibly deploy advanced deep- learning analyses.