Mapping the computational similarity of individual neurons within large-scale ensemble recordings using the SIMNETS...

Area of Science:

• Neuroscience
• Computational Neuroscience
• Systems Neuroscience

Background:

• Large-scale neural recordings offer unprecedented insight into brain activity at single neuron resolution.
• Increasing data complexity necessitates advanced analytical methods beyond traditional approaches.
• Analyzing multi-scale cortical network dynamics requires scalable and efficient computational tools.

Purpose of the Study:

• To introduce the Similarity Networks (SIMNETS) analysis framework for large-scale neural data.
• To enable the identification and visualization of neuronal groups with similar computational functions.
• To provide a method for comparing information processing across neuronal populations.

Main Methods:

• Developed the SIMNETS pipeline for embedding neurons into low-dimensional maps based on spike train similarity.
• Utilized pairwise spike train similarity (SSIM) matrices to capture intrinsic neuronal relationships across experimental conditions.
• Employed three public neural population datasets (visual, motor, hippocampal CA1) for validation.

Main Results:

• Demonstrated SIMNETS' ability to identify putative subnetworks (clusters of neurons with similar computational properties).
• Validated the framework's efficiency and scalability on diverse neural datasets.
• Implemented a novel statistical test to assess the significance of detected neuron clusters.

Conclusions:

• SIMNETS offers an efficient and scalable approach to analyze complex neural population data.
• The framework facilitates the rapid examination of neuronal network computational structure at multiple scales.
• SIMNETS leverages the intrinsic properties of single-unit spike trains to uncover functional network organization.