SNNUtils models

Heterogeneous short-term plasticity parameters sampled from Tsodyks-Markram fits of mouse cortical connections (Allen Institute synaptic physiology; Campagnola et al. 2022). sample_stp_params samples the whole table, sample_stp_campagnola one connection class (:exc_exc, :exc_inh, :inh_exc, :inh_inh).

Output	Column	Units	Meaning
`τF`	`fit_tau_fac`	ms	facilitation time constant
`τD`	`fit_tau_rec`	ms	recovery (depression) time constant
`U`	`fit_U`	-	utilisation of synaptic efficacy
`w`	`fit_w`	as in the table	fitted amplitude (only with `weights = true`)

τD, τF, U are the fields of MarkramSTPParameterHet (Plasticity rules).

using SpikingNeuralNetworks, SNNUtils
p = sample_stp_campagnola(100, :exc_exc)          # τF, τD, U, w for 100 samples
q = sample_stp_params(100)                        # whole table, without weights
stp = SNN.MarkramSTPParameterHet(τD = Float32.(q.τD), τF = Float32.(q.τF), U = Float32.(q.U))

SNNUtils.sample_stp_campagnola — Method

sample_stp_campagnola(N, type; df = stp_data, weights = true)

Like sample_stp_params but restricted to one connection class: type is one of :exc_exc, :exc_inh, :inh_exc, :inh_inh (presynaptic class, postsynaptic class). Returns (τF, τD, U, w) by default (weights = true). The argument df is ignored.

The name refers to Campagnola et al. (2022, Science), the Allen Institute synaptic-physiology survey of mouse and human neocortex, which is the likely source of the bundled fit table; the source file does not state the reference.

Example

using SNNUtils
p = sample_stp_campagnola(5, :exc_inh)

source

SNNUtils.sample_stp_params — Method

sample_stp_params(N; df = stp_data, weights = false)

Draw N sets of Tsodyks-Markram short-term plasticity parameters from the table df of fitted synapses (default: the whole filtered table SNNUtils.stp_data).

For each sample, every parameter is taken from an independently drawn random row (parameters are therefore not jointly sampled from the same synapse).

Returns

(τF, τD, U) (and w if weights = true): vectors of length N with the facilitation time constant (ms, column fit_tau_fac), the depression/recovery time constant (ms, fit_tau_rec), the utilisation U (fit_U) and the fitted amplitude (fit_w, units of the source table). The vectors are Float64; τD, τF, U correspond to the (Float32 vector) fields of SNNModels' MarkramSTPParameterHet.

Example

using SNNUtils
p = sample_stp_params(10)
p.τD

source