DataJoint SciOps — Neuropixels

Written By Data Joint

Neuropixels probes

Neuropixels electrode arrays are rapidly gaining momentum as an instrument of choice for multicellular electrophysiology. These probes allow simultaneous recordings from hundreds or thousands of cells in awake behaving animal subjects across multiple brain areas. The recordings are commonly combined with other modalities: sensory stimulation, behavior recordings, brain state monitoring, histology, and others. A regular annual training course covers the basics of using the probes and analyzing the data. It takes a considerable effort and resources to establish data collection and analysis in a lab.

DataJoint SciOps for Neuropixels

To save time and effort for many neuroscience teams, DataJoint offers a cloud-based workflow for analyzing Neuropixels recordings — hosted, run, and monitored by our expert team. With this service, scientists will simply upload the raw data and will be able to review, access, and download the spiking signals through a cloud-based Jupyter notebook.

The solution is based on the NIH U24-funded DataJoint Elements project in collaboration with leading neuroscience labs to compile and disseminate proven experiment workflows.

What’s included

DataJoint workflows assemble data entry, collection, and analysis into an automated pipeline. The baseline pipeline for Neuropixels recordings includes:

  • DataJoint LabBook: an online user interface for managing information about animals (housing, genotyping, procedures), instruments, and recording sessions.

  • Raw data upload and management: to complete a recording session, the user must run an upload utility that copies the raw data to the SciOps servers for processing and long-term storage. The subscription includes five years of raw data storage.

  • Automated spike sorting: as soon as the raw data gets uploaded, the spike sorting (KiloSort) utility is launched. Upon completion, an email notification will be sent.

  • Online visualization and NWB export: Each workflow comes with an online Jupyter environment with notebooks for querying, visualization, and downloading the processed data. The scripts also allow selecting and exporting the data into a collection of Neurodata Without Borders files for sharing. Data may be queried efficiently across all stored sessions.

  • Transparency and reproducibility: Each workflow comes with a dedicated GitHub repository specifying the code and the scripts used for the data analysis. This includes database schemas, the computing environment, dependencies, and infrastructure to enable the research team to reproduce the entire process with their own resources. The entire datasets, including raw and processed data are made available for export.

  • Support: Our expert team of neural data engineers is available to resolve technical issues in pipeline operation.

What’s coming next

DataJoint Elements aims to provide standardized workflows for the major modalities of neurophysiology experiments. SciOps workflows are continually improved and extended with updated versions of processing algorithms and downstream analysis and visualizations. Our team works closely with tool developers to integrate latest improvements. Workflows can be extended to multimodal experiments synchronizing multiple modalities and adding new analysis steps.

The following features are planned for addition to the Neuropixels SciOps service in the short term:

  • Optional integration of manual spike sorting curation (with the curation performed offline)

  • Quality Control metrics for spike sorting

  • Support for trials events and PSTH plots

  • Near-lossless raw data compression to accelerate the upload

Benefits of SciOps

  • Time to launch: start experiments sooner by reducing the time to configure the data analysis.

  • Simplified computing resources: Data analysis tools require diverse hardware configurations. DataJoint SciOps provides the required computing infrastructure dynamically for each type of analysis.

  • Reducing effort: after each experiment, simply upload the data and it automatically triggers the downstream analysis steps.

  • Staying current: DataJoint SciOps will incorporate validated upgrades from the tool developers.

  • Keeping organized: the recorded sessions are always neatly organized in a relational database for queries and analysis.

  • Extensibility: DataJoint workflows can be extended with new automated analyses.

  • Collaborating and publishing: sharing data with collaborators or submitting data to publishing portals or data archives no longer needs to become a project in itself and is streamlined by DataJoint SciOps.

Learn more

  • Contact us for a demo for your experimental setup.

  • Find DataJoint at SfN 2021 in Chicago.

  • The 2021 Neuropixels Workshop at UCL will include a brief demonstration of DataJoint SciOps.

Data Joint
Previous

University of Bonn Project
Next

NYU Training Videos on YouTube