Transcriptomic cell type structures in vivo neuronal activity across multiple time scales

Abstract

Cell type is hypothesized to be a key determinant of the role of a neuron within a circuit. However, it is unknown whether a neuron's transcriptomic type influences the timing of its activity in the intact brain. In other words, can transcriptomic cell type be extracted from the time series of a neuron's activity? To address this question, we developed a new deep learning architecture that learns features of interevent intervals across multiple timescales (milliseconds to >30 min). We show that transcriptomic cell class information is robustly embedded in the timing of single neuron activity recorded in the intact brain of behaving animals (calcium imaging and extracellular electrophysiology), as well as in a bio-realistic model of visual cortex. In contrast, we were unable to reliably extract cell identity from summary measures of rate, variance, and interevent interval statistics. We applied our analyses to the question of whether transcriptomic subtypes of excitatory neurons represent functionally distinct classes. In the calcium imaging dataset, which contains a diverse set of excitatory Cre lines, we found that a subset of excitatory cell types are computationally distinguishable based upon their Cre lines, and that excitatory types can be classified with higher accuracy when considering their cortical layer and projection class. Here we address the fundamental question of whether a neuron, within a complex cortical network, embeds a fingerprint of its transcriptomic identity into its activity. Our results reveal robust computational fingerprints for transcriptomic types and classes across diverse contexts, defined over multiple timescales.

PhysMAP - interpretablein vivoneuronal cell type identification using multi-modal analysis of electrophysiological data

“…Another notable electrophysiological property that differs between cell types is a neuron's intrinsic firing rate pattern called its inter-spike interval (ISI) distribution (Latuske et al, 2015;Schneider et al, 2023).…”

Simulations of cortical networks using spatially extended conductance‐based neuronal models

“…Schneider et al. (2022) used the model, together with experimental datasets, to show that cell types embed a fingerprint of their transcriptomic identity into functional activity across multiple time scales that can be recorded in the context of the whole network. Chen et al.…”

“…Combined with the mesoscopic connectome data (Harris et al, 2019), a version of the model that contains the whole mouse cortex has also been created (Reimann et al, 2019). The Allen Institute mouse V1 model has been used by different groups to study the statistics of cortical connectivity and its possible effects on network computations (Giacopelli et al, 2021;Stöckl et al, 2021), relationships between connectivity and neural dynamics (Chen et al, 2022;Jabri & MacLean, 2022;Schneider et al, 2022), and circuit mechanisms of behavioural responses (Scherr & Maass, 2021). Schneider et al (2022) used the model, together with experimental datasets, to show that cell types embed a fingerprint of their transcriptomic identity into functional activity across multiple time scales that can be recorded in the context of the whole network.…”

“…The Allen Institute mouse V1 model has been used by different groups to study the statistics of cortical connectivity and its possible effects on network computations (Giacopelli et al, 2021;Stöckl et al, 2021), relationships between connectivity and neural dynamics (Chen et al, 2022;Jabri & MacLean, 2022;Schneider et al, 2022), and circuit mechanisms of behavioural responses (Scherr & Maass, 2021). Schneider et al (2022) used the model, together with experimental datasets, to show that cell types embed a fingerprint of their transcriptomic identity into functional activity across multiple time scales that can be recorded in the context of the whole network. Chen et al (2022) trained this model of mouse V1 to perform tasks such as stimulus classification or discrimination, evidence accumulation, etc., and showed that the model is remarkably robust to noise and achieves high accuracy with relatively few learning epochs, distinguishing it from the more standard artificial intelligence (AI) approaches such as Convolutional Neural Networks.…”

Seeing the forest and the tree: Building representations of both individual and collective dynamics with transformers

“…In many cases, studying systems from this population-level perspective has provided important insights into collective computations and emergent behaviors [7]; however, it also tends to obscure the contributions of different individuals' dynamics to the final prediction or inference. This is significant in settings where the dynamics of different individuals are of interest, either due to their different functional roles in the system [8,9,6,10], or due to shift in their dynamics because of sensor displacement or corruption [11][12][13]. Moving forward, we need methods that can build good population-level representations while also providing an interpretable view of the data at the individual level.…”