Data architecture for a large-scale neuroscience collaboration

Abstract

The International Brain laboratory (IBL) is a collaboration aiming to understand the neural basis of decision-making. Ten experimental labs use multiple neural recording modalities in diverse brain structures of mice making perceptual decisions. A primary requirement of IBL is to establish a data architecture that integrates data from all labs and modalities together. We have developed a system that allows users across 5 countries to automatically contribute data and metadata, search for relevant data, and share code for exploratory analysis.To accurately record metadata about subjects and experiments in a searchable and accessible way, we have developed a user-friendly, web-based electronic lab notebook system for colony and experiment management (Alyx). Alyx is a small relational metadatabase that records relevant information about each subject (age, strain, procedures), alongside information about experiments and their resulting data files. Once an experiment is completed, data registered in Alyx is automatically uploaded to a central server via Globus transfer. Users can search and access the data using a lightweight interface called the Open Neurophysiology Environment (ONE), implemented in Python and MATLAB. ONE defines a list of DatasetTypes, as arrays of predetermined shapes and units. Users search the database by running a command that returns an ID identifying experiments matching their criteria. A command one.load returns the requested DatasetTypes as a numerical array, downloading it from the central server and caching on the user's machine to avoid repeated downloads. Additional commands provide further ways to list experiments on the server, and the dataset types they each contain. This system allows users to search, load and process data that was collected in laboratories spanning multiple geographical locations. To visualize these data, we have developed a pipeline for automated analysis, based on the DataJoint framework.

A Primer on Motion Capture with Deep Learning: Principles, Pitfalls, and Perspectives

“…The recent large open dataset from the Allen Institute includes imaging data of six cortical and two thalamic regions in response to various stimuli classes as well as pupil tracking with DeepLab-Cut (Siegle et al, 2019). The International Brain Lab has integrated DeepLabCut into their workflow to track multiple body parts of decision-making mice including their pupils (Harris et al, 2019).…”

“…Essentially, now every lab can train appropriate algorithms for their application and turn videos into accurate measurements of posture. If setups are sufficiently standardized, these algorithms already broadly generalize, even across multiple laboratories as in the case of the International Brain Lab (Harris et al, 2019). But how do we get there, and how do we make sure the needs of animal pose estimation for neuroscience applications are met?…”

Climate crisis and ecological emergency: Why they concern (neuro)scientists, and what we can do

“…Scientific computing is another surprisingly resource-intensive enterprise. Neuroscientists require ever-increasing compute resources to analyse ever-growing datasets ( Glasser et al, 2016 ; Littlejohns et al, 2020 ; The International Brain Laboratory et al, 2019 ), and this brings with it increased energy demands and carbon costs. Data centres and IT equipment are environmentally costly to build and energy-hungry to run (in part due to the requirement for constant air conditioning, even when data are not being analysed).…”

Knowledge across networks: how to build a global neuroscience collaboration

“…While physical proximity between collaborating researchers has its advantages, IBL's distributed model allows team composition to be adjusted more flexibly and nimbly, drawing together talents from around the world. Indeed, our own platform papers [2,3] highlight the sheer diversity of researchers needed to complete these large-scale projects. Expanding researcher roles beyond trainees should be a central goal if our field wishes to stabilize large-scale collaborative science in traditional academia [30]; it is promising to consider the new types of researchers we can attract if we don't need to bring them into a single lab, but rather can bring the lab directly to them.…”

“…The collaboration is currently distributed over 22 laboratories across six countries, with each location hosting several members who implement IBL tools and resources within the existing laboratory infrastructure. Our network has unified the efforts of nearly 80 neuroscientists to develop an experimental ecosystem generating high-throughput behavioral data in mice [2], a pipeline for acquiring large-scale, high-density neural recordings during behavior, and a public-facing digital architecture to host these data [3].…”