Modularization and integration of the International Brain Laboratory spike-sorting pipeline into SpikeInterface

PROJECT SUMMARY The parent grant titled “State-dependent Decision-making in Brainwide Neural Circuits” has the goal of understanding how internal states influence decisions and identifying the underlying neural mechanisms. The project builds upon the International Brain Laboratory (IBL) effort, which brings together a large consortium of

laboratories to standardize and parallelize neural data acquisition and analysis. The parent project includes an extensive data acquisition work package leveraging Neuropixels 2.0 probes, which are state-of-the-art high- density neural devices that can record from up to 384 channels simultaneously. The large amount of raw

electrophysiology data (~100 TB) requires the development of cutting-edge analysis tools for neural data science. Specifically, “spike sorting” is the crucial processing step that enables the extraction of single-neuron activity from the recorded signals; the sorted single-neuron signals are then used for downstream analysis and

modeling. In this direction, the Data Science Core of the parent project has been developing computational pipelines and novel processing steps to automate and improve the outcome of spike sorting for the IBL data. However, these novel tools are tightly bound to the IBL analysis pipeline and would benefit from a generalization

and standardization effort to be more useful to the general neurophysiology community. The main goal of this supplement project is therefore to integrate such state-of-the-art tools into the well- established SpikeInterface software framework1 in order to maximize outreach and accessibility to the broad

electrophysiology community. In doing so, we aim to modularize the developed pipeline into discrete, decoupled, and interchangeable processing sorting components and to improve their software implementation in terms of speed, efficiency, and scalability, as well as documentation and testing. Finally, we will build a software

infrastructure to enable users to construct and run full spike sorting pipelines into containerized and cloud-ready solutions.