On the effect of neuronal spatial subsampling in small-world networks

Area of Science:

• Neuroscience
• Network Science
• Computational Biology

Background:

• Real-world neuronal network analysis is inherently limited by sampling, potentially biasing functional interpretations.
• Understanding how network properties derived from subsamples relate to the global network is crucial for accurate analysis.

Purpose of the Study:

• To investigate the scaling properties of network descriptors (average clustering coefficient, path length, small-world propensity) under spatial sampling.
• To determine if information from network subsamples can be extrapolated to predict global network topology and function.

Main Methods:

• Spatial sampling was applied to small-world networks, mimicking physical neighbor measurements used in neuronal network studies.
• Analysis was performed on both in silico and in vivo data to assess the generalizability of findings.
• Mathematical manipulations were used to reduce topology dependence during the scaling analysis.

Main Results:

• Network properties (CC, PL, SWP) were found to scale with both the size of the sampled network and the global network topology.
• Qualitative behaviors of descriptors were consistent across analyzed networks, enabling approximate predictions of global graph properties from subgraph information.
• Below a critical spatial sampling threshold, subgraphs lacked sufficient information for reliable extrapolation to the global network.

Conclusions:

• The size of sampled subgraphs is a critical factor for the accurate extrapolation of findings to the entire neuronal network.
• Network analysis requires careful consideration of sampling size to ensure reliable functional interpretations.
• This study provides a framework for understanding the limitations and possibilities of network analysis based on partial data.