Supercomputer framework for reverse engineering firing patterns of neuron populations to identify their synaptic inputs

• DOI: 10.1101/2022.12.09.519818
• Publication Date: 2022-12-11T23:20:14Z
• Authors: Matthieu K. Chardon; Y. Curtis Wang; Marta Garcia; Emre Besler; J. Andrew Beauchamp; Michael D’Mello; Randall K. Powers; Charles J. Heckman
• Keywords: Neuromodulation

Abstract In this study, we develop new reverse engineering (RE) techniques to identify the organization of synaptic inputs generating firing patterns populations neurons. We tested these in silico allow rigorous evaluation their effectiveness, using remarkably extensive parameter searches enabled by massively-parallel computation on supercomputers. chose spinal motoneurons as our target neural system, since process all motor commands and have well established input-output properties. One set simulated was driven 300,000+ combinations excitatory, inhibitory, neuromodulatory inputs. Our goal determine if had sufficient information RE identification input combinations. Like other systems, motoneuron system is likely non-unique. This non-uniqueness could potentially limit approach, many can produce similar outputs. However, simulations revealed that contained sharply restrict solution space. Thus, approach successfully generated estimates actual excitation, inhibition, neuromodulation, with variances accounted for ranging from 75% 90%. It striking nonlinearities induced neuromodulation did not impede RE, but instead distinctive features aided RE. These demonstrate potential form analysis. ever-increasing capacity supercomputers will increasingly accurate neuron systems.