Parametric computation and persistent gamma in a cortical model (Chambers et al. 2012)

 Download zip file 
Help downloading and running models
Using the Traub et al (2005) model of the cortex we determined how 33 synaptic strength parameters control gamma oscillations. We used fractional factorial design to reduce the number of runs required to 4096. We found an expected multiplicative interaction between parameters.
1 . Chambers JD, Bethwaite B, Diamond NT, Peachey T, Abramson D, Petrou S, Thomas EA (2012) Parametric computation predicts a multiplicative interaction between synaptic strength parameters that control gamma oscillations. Front Comput Neurosci 6:53 [PubMed]
Model Information (Click on a link to find other models with that property)
Model Type: Realistic Network; Axon; Synapse; Channel/Receptor; Dendrite;
Brain Region(s)/Organism:
Cell Type(s): Neocortex L5/6 pyramidal GLU cell; Neocortex L2/3 pyramidal GLU cell; Neocortex V1 interneuron basket PV GABA cell; Neocortex fast spiking (FS) interneuron; Neocortex spiny stellate cell; Neocortex spiking regular (RS) neuron; Neocortex spiking low threshold (LTS) neuron;
Channel(s): I A; I K; I K,leak; I K,Ca; I Calcium; I_K,Na;
Gap Junctions: Gap junctions;
Receptor(s): GabaA; AMPA; NMDA;
Transmitter(s): Gaba; Glutamate;
Simulation Environment: NEURON;
Model Concept(s): Oscillations; Parameter sensitivity;
Implementer(s): Thomas, Evan [evan at]; Chambers, Jordan [jordandchambers at];
Search NeuronDB for information about:  Neocortex L5/6 pyramidal GLU cell; Neocortex L2/3 pyramidal GLU cell; Neocortex V1 interneuron basket PV GABA cell; GabaA; AMPA; NMDA; I A; I K; I K,leak; I K,Ca; I Calcium; I_K,Na; Gaba; Glutamate;
balcomp.hoc *
defvar.hoc *
lbcreate.hoc *
mscreate.hoc *
traubcon.hoc *
traubcon_net.hoc *

objref gidvec, cvec, splitxlist, splitixlist, cpu, splitcplx, splitindex, bal
objref splitbres, splitbrlist
bal = new LoadBalance()

gidvec = new Vector()
cvec = new Vector()
splitxlist = new List()
splitixlist = new List()
splitbrlist = new List()
cpu = new Vector()
splitcplx = new Vector()
splitindex = new Vector()
splitbres = new Vector()

proc rdat() {local i, j, k, n1, n2, n3, c, n3 localobj f, vecx, vecix, vecb
	f = new File()
	n1 = f.scanvar()
	for i=0, n1-1 {
		n2 = f.scanvar()
		for j=0, n2-1 {
			gid = f.scanvar()
			c = f.scanvar()
			n3 = f.scanvar()
			vecx = new Vector(n3)
			vecix = new Vector(n3)
			vecb = new Vector(n3)
			for k = 0, n3-1 {
				vecx.x[k] = f.scanvar()
				vecix.x[k] = f.scanvar()
				vecb.x[k] = f.scanvar()

func balance() {local i, err  localobj f, si, s
	s =new String()
	sprint(s.s, "%s.dat", $s2)
	err = bal.distrib($1, cvec, splitxlist, splitixlist, \
		cpu, splitcplx, splitindex, splitbrlist, splitbres)
	f = new File()
	sprint(s.s, "%s.%d", $s2, $1)
	f.printf("%d\n", gidvec.size)
	si = cpu.sortindex
//print gidvec.size, si.size, cpu.size, gidvec.size, splitindex.size
//print splitbres, splitbres.size
	for i=0, gidvec.size-1 {
		f.printf("%d %d %d %d %d %d\n", cpu.x[si.x[i]], \
			gidvec.x[si.x[i]], splitindex.x[si.x[i]], \
			splitbres.x[si.x[i]], \
			splitcplx.x[si.x[i]], cvec.x[si.x[i]], \
			bal.cplx.x[cpu.x[si.x[i]]] )
	return err

objref pc
pc = new ParallelContext()
ncpu = 32*2^(
percenterr = balance(ncpu, "splitbal")
//print cvec.sum
//print bal.cplx.sum
{printf("ncpu=%d load balance error = %d%% average load = %d   max load = %d\n", ncpu, percenterr, cvec.sum/ncpu, bal.cplx.max)}

proc chkbal() {local i localobj ix, split
	ix = new Vector()
	split= new Vector()
	for i=0, cpu.max-1 {
		ix.indvwhere(cpu, "==", i)
		split.index(splitindex, ix)
		if (split.indvwhere("!=", 0).size > 1) {
			printf("cpu %d with %d\n", i, split.size)

Loading data, please wait...