Decoding neuron identity from spike recordings using AI and ion channel theory

What is it about?

Neurons communicate by firing brief electrical pulses called spikes. The precise timing of these spikes depends on the mix of ion channels in the neuron's membrane, but figuring out which ion channels are present just by watching spikes is extremely difficult. To make things harder, many different ion channel combinations can produce nearly identical spiking patterns, a phenomenon called neuronal degeneracy. We developed a two-step computational method to tackle this problem. First, a lightweight artificial neural network reads a spike recording and predicts three numbers, called Dynamic Input Conductances (DICs), that compactly summarize what kind of ion channel activity is driving the neuron. Second, an algorithm uses those three numbers to generate hundreds of realistic neuron models, each with a different ion channel composition, that all reproduce the original spike pattern. The whole process runs in milliseconds on a standard laptop. We tested the method on two different neuron types spanning regular spiking, bursting, and slow pacemaking, and it worked accurately in all cases, even when the spike recordings were noisy. We also release the method as free, open-source software with a graphical interface, so researchers without coding experience can use it directly from their recordings.

Featured Image

Photo by Milad Fakurian on Unsplash

Photo by Milad Fakurian on Unsplash

Why is it important?

Understanding which ion channels drive a neuron's activity is fundamental to neuroscience, yet it has remained out of reach for most experimental settings. Current methods either require invasive voltage recordings, work only for a single solution rather than capturing the full range of equivalent models, or produce results that are hard to interpret biologically. This work is the first to combine deep learning with a principled dynamical systems framework to rapidly reconstruct not just one, but entire populations of biologically plausible neuron models directly from spike times. The interpretable intermediate representation (DICs) means researchers can understand why certain ion channel combinations produce similar activity, not just that they do. The speed and accessibility of the approach, combined with the open-source graphical interface, make it practical for experimental laboratories. This opens direct applications in neuromodulation studies, pharmacology, deep brain stimulation, and the design of neuromorphic hardware.

Perspectives