Bayesian methods for event analysis of intracellular currents

Abstract

Background: Investigation of neural circuit functioning often requires statistical interpretation of events in subthreshold electrophysiological recordings. This problem is non-trivial because recordings may have moderate levels of structured noise and events may have distinct kinetics. In addition, novel experimental designs that combine optical and electrophysiological methods will depend upon statistical tools that combine multimodal data.New method: We present a Bayesian approach for inferring the timing, strength, and kinetics of postsynaptic currents (PSCs) from voltage-clamp electrophysiological recordings on a per event basis. The simple generative model for a single voltage-clamp recording flexibly extends to include additional structure to enable experiments designed to probe synaptic connectivity. Results:We validate the approach on simulated and real data. We also demonstrate that extensions of the basic PSC detection algorithm can handle recordings contaminated with optically evoked currents, and we simulate a scenario in which calcium imaging observations, available for a subset of neurons, can be fused with electrophysiological data to achieve higher temporal resolution.Comparison with existing methods: We apply this approach to simulated and real ground truth data to demonstrate its higher sensitivity in detecting small signal-to-noise events and its increased robustness to noise compared to standard methods for detecting PSCs. Conclusions:The new Bayesian event analysis approach for electrophysiological recordings should allow for better estimation of physiological parameters under more variable conditions and help support new experimental designs for circuit mapping.

Complementary networks of cortical somatostatin interneurons enforce layer specific control

“…If individual SST cells target either L4 or L5 PCs, but not both, then we should never observe common input to pairs of L4 and L5 PCs when photo-stimulating single SST neurons. This can be tested by mapping optogenetically evoked unitary SST inhibitory connections onto multiple PCs recorded simultaneously and analyzing the spatiotemporal coincidence of evoked IPSCs onto different pairs of PCs, thereby measuring the amount of common input shared between pairs of PCs in different layers (Yoshimura et al, 2005; Morgenstern et al, 2016), (Merel et al, 2016). Although this approach does not discriminate between MCs and NMCs directly, it performs a more stringent test by extending our hypothesis to apply to the structure of the outputs of the L5 SST population as a whole, rather than the sparser subsets set labeled in the GFP lines.…”

Complementary networks of cortical somatostatin interneurons enforce layer specific control

“…In order to conduct comprehensive mapping of individual SOM outputs, we developed a novel approach to map neural circuits at high spatiotemporal resolution using two photon optogenetics and a statistical pipeline for detecting synaptic connections (Merel et al, 2016) and evoked inhibitory postsynaptic current (IPSC) synchrony. In addition, we employed a somatargeted opsin (Wu et al, 2013), which has the advantage of providing far superior spatial resolution during photo-stimulation than non-targeted opsins (Fig.…”

“…To determine which locations evoked responses in the voltage-clamp recordings, first we detected IPSCs using a Bayesian modeling approach via Gibbs sampling (Merel et al, 2016). Event times were estimated by first binning all of the time samples at 1 ms resolution and then finding maxima of those timeseries (using findpeaks in MATLAB).…”

MOD: A novel machine-learning optimal-filtering method for accurate and efficient detection of subthreshold synaptic eventsin vivo

“…In conclusion, these results indicate that MOD is able to dissect temporal changes in event frequency and peak amplitude, as required for a mechanistic analysis of single-neuron computations under in vivo conditions. (Jonas et al, 1993;Clements and Bekkers, 1994;Pernía-Andrade et al, 2012;Merel et al, 2016). Third, cross-validation used in the MOD algorithm provides quantitative information about the expected detection errors of unscored data.…”

“…For the simpler but related problem of detection of spontaneous synaptic activity in vitro (Kavalali, 2015), several different methods were proposed. These include amplitude threshold methods, derivative-based methods (Maier et al, 2011), template-fit algorithms (Jonas et al, 1993;Clements and Bekkers, 1997;Chadderton et al, 2004), deconvolution methods (Pernía-Andrade et al, 2012), and Bayesian approaches that consider distributions of templates rather than single templates (Merel et al, 2016). However, these methods cannot be easily applied to in vivo data sets.…”