DRIADA: Python toolkit unifies neural data analysis across scales

DRIADA, short for Dynamic Research Infrastructure for Advanced Data Analysis, is an open-source Python framework designed to tackle one of neuroscience's biggest bottlenecks: the inability to easily move between single-neuron, population, and network-level analyses. Developed by Nikita Pospelov and nine colleagues at the Kurchatov Institute, the toolkit introduces a unified data model that integrates neural signals (calcium imaging, spike trains) with time-aligned behavioral data. This allows researchers to perform selectivity testing, dimensionality reduction, and network analysis within a single, coherent workflow — something existing tools struggle to do because they are typically built for one paradigm.

The team validated DRIADA on multiple datasets. In hippocampal calcium imaging from 13 mice exploring an open field, selectivity-based filtering restored a collapsed spatial embedding, showing the toolkit can uncover hidden neural representations. A reverse analysis revealed that ~57% of neurons that were informative about the leading manifold dimensions showed no selectivity to any of 11 behavioral features tested — a finding that challenges assumptions about how behavior maps onto neural activity. On a simulated toroidal attractor network benchmark, four independent DRIADA modules each correctly recovered the expected topology. The framework supports calcium imaging, spike trains, and simulated networks, making cross-scale analysis routine for the first time.