# scalemutations.py
# A script for scaling down the effects of genetic variants displayed in the list MT returned by mutation_stuff.getMT()
# For each variant MT[i][j], a coefficient c is looked for which fulfils the following condition: If the variant is
# scaled with c-eps for small eps > 0, then the scaled variant neuron obeys the conditions I-V (it behaves similarly to
# the control neuron). If the variant is scaled with c+eps for small eps, then the scaled variant neuron does not obey
# some of the conditions I-V (it behaves in a different manner than the control neuron). If c is larger than 1.99, then
# such a factor c was not found on the range [0,2], meaning that the variant need not be downscaled. The resulting scaling
# factors are saved to files 'scalings_cs'+str(icell)+'_'+str(counter)+'.sav', where icell refers to the cell morphology
# and counter refers to the variant number. Note that if variant MT[i][j] reports ranges of different parameter changes,
# then the scaling is performed for each combination of the end points of these ranges.
#
# The input code for the hoc-interface is based on BAC_firing.hoc by Etay Hay (2011)
#
# Tuomo Maki-Marttunen, Jan 2015
# (CC BY)
from neuron import h
import mytools
from pylab import *
import pickle
import sys
import mutation_stuff
# Get the table of variants and the default values for the model parameters
MT = mutation_stuff.getMT()
defVals = mutation_stuff.getdefvals()
keyList = defVals.keys()
for idefval in range(0,len(keyList)):
if type(defVals[keyList[idefval]]) is not list:
defVals[keyList[idefval]] = [defVals[keyList[idefval]], defVals[keyList[idefval]]] #make the dictionary values [somatic, apical]
updatedVars = ['somatic','apical','basal'] # the possible classes of segments that defVals may apply to
whichDefVal = [0,1,0] # use the defVal[0] for somatic and basal segments and defVal[1] for apical segments
# Get the properties of a control neuron
unpicklefile = open('control.sav', 'r')
unpickledlist = pickle.load(unpicklefile)
unpicklefile.close()
spikfreqs_control_All = unpickledlist[0] #List of spiking frequencies
timesc_control_All = unpickledlist[1] #Times for limit cycle
Vsomac_control_All = unpickledlist[2] #V for limit cycle
VDerivc_control_All = unpickledlist[3] #Derivative for limit cycle
VDcoeff_control_All = unpickledlist[4] #Coefficient for the derivatives
Is_control = unpickledlist[19] #List of current amplitudes used for calculating spikfreqs_control_All
Is = [0.2,0.4,0.6,0.8,1.0,1.2,1.4]
theseCoeffsAllAll = [] # Collect here the scaling coefficients
#The somatic current amplitudes [A1,A2,A3] for both cell morphologies in three conditions: Somatic step alone (A1), Synaptic alone (A2), and Somatic+Synaptic (A3):
squareAmps = [[0.696,0,1.137],[0.872,0,0.993]]
#The synaptic current amplitudes [S1,S2,S3] for both cell morphologies in three conditions: Somatic step alone (S1), Synaptic alone (S2), and Somatic+Synaptic (S3):
epsp_gmaxs = [[0,0.0612,0.100],[0,0.455,0.518]]
for icell in range(0,2): # icell goes through the two cell morphologies (cell #3 wouldn't produce spikes)
spikfreqs_control = mytools.interpolate(Is_control,spikfreqs_control_All[icell],Is)
Vsomac_control = Vsomac_control_All[icell]
VDerivc_control = VDerivc_control_All[icell]
VDcoeff_control = VDcoeff_control_All[icell]
timesc_control = timesc_control_All[icell]
print spikfreqs_control
morphology_file = "morphologies/cell"+str(icell+1)+".asc"
biophys_file = "models/L5PCbiophys3.hoc"
template_file = "models/L5PCtemplate.hoc"
v0 = -80
ca0 = 0.0001
proximalpoint = 400
distalpoint = 620
BACdt = 5.0
#Initialize the model
h("""
load_file("stdlib.hoc")
load_file("stdrun.hoc")
objref cvode
cvode = new CVode()
cvode.active(1)
load_file("import3d.hoc")
objref L5PC
load_file(\""""+biophys_file+"""\")
load_file(\""""+template_file+"""\")
L5PC = new L5PCtemplate(\""""+morphology_file+"""\")
access L5PC.soma
objref st1
st1 = new IClamp(0.5)
L5PC.soma st1
L5PC.distribute_channels("apic","gIhbar_Ih",2,-0.8696,3.6161,0.0,1.0*2.0870,0.0002)
L5PC.distribute_channels("apic","gCa_HVAbar_Ca_HVA",3,1.0,0.1,685.0,885.0,1.0*0.000555)
L5PC.distribute_channels("apic","gCa_LVAstbar_Ca_LVAst",3,1.0,0.01,685.0,885.0,1.0*0.0187)
objref sl,st2,ns,syn1,con1,isyn, tvec
isyn = new Vector()
tvec = new Vector()
sl = new List()
double siteVec[2]
sl = L5PC.locateSites("apic","""+str(distalpoint)+""")
maxdiam = 0
for(i=0;i<sl.count();i+=1){
dd1 = sl.o[i].x[1]
dd = L5PC.apic[sl.o[i].x[0]].diam(dd1)
if (dd > maxdiam) {
j = i
maxdiam = dd
}
}
siteVec[0] = sl.o[j].x[0]
siteVec[1] = sl.o[j].x[1]
print "distalpoint gCa_HVA: ", L5PC.apic[siteVec[0]].gCa_HVAbar_Ca_HVA
print "distalpoint gCa_LVA: ", L5PC.apic[siteVec[0]].gCa_LVAstbar_Ca_LVAst
L5PC.apic[siteVec[0]] st2 = new IClamp(siteVec[1])
st2.amp = 0
L5PC.apic[siteVec[0]] {
st2
syn1 = new AlphaSynapse(siteVec[1])
syn1.e = 0
syn1.tau = 5
syn1.onset = 200 + """+str(BACdt)+"""
cvode.record(&syn1.i,isyn,tvec)
}
objref vsoma, vdend, recSite, vdend2, isoma, cadend, cadend2, casoma
vsoma = new Vector()
casoma = new Vector()
vdend = new Vector()
cadend = new Vector()
vdend2 = new Vector()
cadend2 = new Vector()
L5PC.soma cvode.record(&v(0.5),vsoma,tvec)
L5PC.soma cvode.record(&cai(0.5),casoma,tvec)
L5PC.apic[siteVec[0]] cvode.record(&v(siteVec[1]),vdend,tvec)
L5PC.apic[siteVec[0]] cvode.record(&cai(siteVec[1]),cadend,tvec)
sl = new List()
sl = L5PC.locateSites("apic","""+str(proximalpoint)+""")
maxdiam = 0
for(i=0;i<sl.count();i+=1){
dd1 = sl.o[i].x[1]
dd = L5PC.apic[sl.o[i].x[0]].diam(dd1)
if (dd > maxdiam) {
j = i
maxdiam = dd
}
}
siteVec[0] = sl.o[j].x[0]
siteVec[1] = sl.o[j].x[1]
L5PC.apic[siteVec[0]] recSite = new IClamp(siteVec[1])
recSite.amp = 0
L5PC.apic[siteVec[0]] {
recSite
}
L5PC.apic[siteVec[0]] cvode.record(&v(siteVec[1]),vdend2,tvec)
L5PC.apic[siteVec[0]] cvode.record(&cai(siteVec[1]),cadend2,tvec)
L5PC.soma isoma = new Vector()
L5PC.soma cvode.record(&st1.i,isoma,tvec)
""")
ITERS = 20
theseCoeffsAll = [] #Save here the downscaling coefficients
theseMutValsAll = [] #Save here the changed parameter values
theseMutVarsAll = [] #Save here the names of changed parameters
counter = -1
for igene in range(0,len(MT)): # Go through the genes
theseCoeffsGene = []
for imut in range(0,len(MT[igene])): # Go through the variants of this gene
theseCoeffsMut = []
nVals = len(MT[igene][imut])*[0]
thesemutvars = []
for imutvar in range(0,len(MT[igene][imut])): # Go through the parameters changed by the variants
thesemutvars.append(MT[igene][imut][imutvar][0])
if type(MT[igene][imut][imutvar][1]) is int or type(MT[igene][imut][imutvar][1]) is float:
MT[igene][imut][imutvar][1] = [MT[igene][imut][imutvar][1]] # If the type of the entry is not list, make it a list of one entry
nVals[imutvar] = len(MT[igene][imut][imutvar][1]) # nVals shows how many end points there are for each parameter variation
cumprodnVals = cumprod(nVals) # take the cumulative product to help determining the combinations of end points
allmutvars = cumprodnVals[len(MT[igene][imut])-1]*[thesemutvars[:]]
allmutvals = []
for iallmutval in range(0,cumprodnVals[len(MT[igene][imut])-1]): # Reserve memory for allmutvals (the values for parameter value change of each combination of end points)
allmutvals.append([0]*len(thesemutvars))
for iallmutval in range(0,cumprodnVals[len(MT[igene][imut])-1]): # Set the values for parameter value change of each combination of end points
for imutvar in range(0,len(MT[igene][imut])):
if imutvar==0:
allmutvals[iallmutval][imutvar] = MT[igene][imut][imutvar][1][iallmutval%nVals[imutvar]]
else:
allmutvals[iallmutval][imutvar] = MT[igene][imut][imutvar][1][(iallmutval/cumprodnVals[imutvar-1])%nVals[imutvar]]
theseMutValsAll.append(allmutvals[:])
theseMutVarsAll.append(allmutvars[:])
for iallmutval in range(0,cumprodnVals[len(MT[igene][imut])-1]): # Go through all combinations of end points of the parameter ranges corresponding to this variant
# Allow parallelization if possible. The variants go from 0 to 51; the index to be considered can be given as the first argument, otherwise, all variants are considered one after another
counter = counter + 1
if len(sys.argv) > 1 and int(float(sys.argv[1])) != counter:
continue
# Use bisection method to find a threshold c
nextCoeffs = [0.0,2.0,1.0] # nextCoeffs[0] shows the lower bound for c, nextCoeffs[1] the upper bound, and nextCoeffs[2] the next one to iterate
for iter in range(0,ITERS+2): # Iterate ITERS times to find the threshold factor c when the downscaled variant violates some of the conditions I-V
thisCoeff = nextCoeffs[min(iter,2)] #First try 0.0 and 2.0, and then start iterating
mutText = "" # Save here a compact string presenting the effects of the scaled variant
# Apply the scaling c for all parameter changes presented by this variant
for imutvar in range(0,len(MT[igene][imut])): # Go through all model parameters
if imutvar > 0 and imutvar%2==0:
mutText = mutText+"\n"
mutvars = allmutvars[iallmutval][imutvar]
if type(mutvars) is str:
mutvars = [mutvars]
mutText = mutText + str(mutvars) + ": "
mutvals = allmutvals[iallmutval][imutvar]
for kmutvar in range(0,len(mutvars)): # Go through the parameters that should be changed at once
if mutvars[kmutvar].find('offm') > -1 or mutvars[kmutvar].find('offh') > -1 or mutvars[kmutvar].find('ehcn') > -1: # Apply linear change to offset and reversal potentials
newVal = [x+thisCoeff*mutvals for x in defVals[mutvars[kmutvar]]] # Add the scaled increment to the default value
if mutvals >= 0 and kmutvar==0:
mutText = mutText + "+" + str(mutvals*thisCoeff) +" mV"
elif kmutvar==0:
mutText = mutText + str(mutvals*thisCoeff) +" mV"
else: # Apply logarithmic change to all other types of model parameters
newVal = [x*(mutvals**thisCoeff) for x in defVals[mutvars[kmutvar]]] # Multiply the default value with a scaled factor
if kmutvar==0:
mutText = mutText + "*" + str(mutvals**thisCoeff)
if kmutvar < len(mutvars)-1:
mutText = mutText + ", "
if mutvars[kmutvar].find('_Ih') > -1: # For Ih, apply the changes to both somatic, apical and basal segments
updateThese = [1,1,1]
elif mutvars[kmutvar].find('_Ca_HVA') > -1 or mutvars[kmutvar].find('_Ca_LVAst') > -1 or mutvars[kmutvar].find('_SKv3.1') > -1 or mutvars[kmutvar].find('_Ca_HVA') > -1 or mutvars[kmutvar].find('_SK_E2') > -1 or mutvars[kmutvar].find('_NaTa_t') > -1 or mutvars[kmutvar].find('_CaDynamics_E2') > -1: # For these channels, apply the change to somatic and apical segments
updateThese = [1,1,0]
elif mutvars[kmutvar].find('_K_Pst') > -1 or mutvars[kmutvar].find('_K_Tst') > -1 or mutvars[kmutvar].find('_Nap_Et2') > -1: # For these channels, apply the change to somatic segments
updateThese = [1,0,0]
elif mutvars[kmutvar].find('_Im') > -1: # For these channels, apply the change to apical segments
updateThese = [0,1,0]
else:
print "Error: str=" + str(mutvars[kmutvar])
updatedVars = [0,0,0]
for iupdated in range(0,3):
if updateThese[iupdated]:
print """forsec L5PC."""+str(updatedVars[iupdated])+""" {
"""+mutvars[kmutvar]+""" = """+str(newVal[whichDefVal[iupdated]])+"""
}"""
h("""forsec L5PC."""+str(updatedVars[iupdated])+""" {
"""+mutvars[kmutvar]+""" = """+str(newVal[whichDefVal[iupdated]])+"""
}""")
print mutText
############################################# Condition 1: Short burst #############################################
tstop = 500.0
squareAmp = squareAmps[icell][0]
squareDur = 150.0
epsp_gmax = 0.0
h("""
tstop = """+str(tstop)+"""
v_init = """+str(v0)+"""
cai0_ca_ion = """+str(ca0)+"""
st1.amp = """+str(squareAmp)+"""
st1.del = 200
st1.dur = """+str(squareDur)+"""
syn1.gmax = """+str(epsp_gmax)+"""
syn1.onset = 200 + """+str(BACdt)+"""
""")
h.init()
h.run()
times=np.array(h.tvec)
Vsoma=np.array(h.vsoma)
spikes = mytools.spike_times(times,Vsoma,-35,-45)
nSpikes1 = len(spikes)
############################################# Condition 2: Distal EPSC #############################################
squareAmp = 0.0
epsp_gmax = epsp_gmaxs[icell][1]
h("""
tstop = """+str(tstop)+"""
v_init = """+str(v0)+"""
cai0_ca_ion = """+str(ca0)+"""
st1.amp = """+str(squareAmp)+"""
st1.dur = """+str(squareDur)+"""
syn1.gmax = """+str(epsp_gmax)+"""
syn1.onset = 200 + """+str(BACdt)+"""
""")
h.init()
h.run()
times=np.array(h.tvec)
Vsoma=np.array(h.vsoma)
spikes = mytools.spike_times(times,Vsoma,-35,-45)
nSpikes2 = len(spikes)
############################################# Condition 3: Somatic stim + EPSC #############################################
squareAmp = squareAmps[icell][2]
squareDur = 10
epsp_gmax = epsp_gmaxs[icell][2]
h("""
tstop = """+str(tstop)+"""
v_init = """+str(v0)+"""
cai0_ca_ion = """+str(ca0)+"""
st1.amp = """+str(squareAmp)+"""
st1.dur = """+str(squareDur)+"""
syn1.gmax = """+str(epsp_gmax)+"""
syn1.onset = 200 + """+str(BACdt)+"""
""")
h.init()
h.run()
times=np.array(h.tvec)
Vsoma=np.array(h.vsoma)
spikes = mytools.spike_times(times,Vsoma,-35,-45)
nSpikes3 = len(spikes)
############################################# Condition 4: IF curve #############################################
spikfreqs = len(Is)*[0]
for iI in range(0,len(Is)):
tstop = 4000.0
squareAmp = Is[iI]
squareDur = 3800.0
epsp_gmax = 0.0
h("""
tstop = """+str(tstop)+"""
v_init = """+str(v0)+"""
cai0_ca_ion = """+str(ca0)+"""
st1.amp = """+str(squareAmp)+"""
st1.del = 200
st1.dur = """+str(squareDur)+"""
syn1.gmax = """+str(epsp_gmax)+"""
syn1.onset = 200 + """+str(BACdt)+"""
""")
h.init()
h.run()
times=np.array(h.tvec)
Vsoma=np.array(h.vsoma)
spikes = mytools.spike_times(times,Vsoma,-35,100)
spikfreqs[iI] = sum([1 for x in spikes if x >= 500.0])/3.5
if iI==4: # use the memb. pot. time course of 1.0nA for the limit cycle
times_lc = times[:]
Vsoma_lc = Vsoma[:]
spikes_lc = spikes[:]
spikfreqdiffsum = sum([abs(x-y) for x,y in zip(spikfreqs,spikfreqs_control)])
spikfreqdiffrel = spikfreqdiffsum/sum(spikfreqs_control)
############################################# Condition 5: Limit cycle #############################################
if len(spikes_lc) < 3:
lcdiff = 1e6
else:
spts = spikes_lc[len(spikes_lc)-3:len(spikes_lc)]
istart = next((i for i,x in enumerate(times_lc) if x > spts[0]))
iend = next((i for i,x in enumerate(times_lc) if x > spts[1]))+4
nsteps = iend-istart-1
Vsomac = Vsoma_lc[istart:iend]
timesc = times_lc[istart:iend]
VDerivc = mytools.membpotderivs(timesc,Vsomac)
VDcoeff = mytools.limitcyclescaledv(Vsomac,VDerivc,Vsomac,VDerivc)
lcdiff1 = mytools.limitcyclediff(Vsomac[1:nsteps-1],VDerivc,Vsomac_control,VDerivc_control,VDcoeff_control)
lcdiff2 = mytools.limitcyclediff(Vsomac_control,VDerivc_control,Vsomac[1:nsteps-1],VDerivc,VDcoeff_control)
lcdiff = 0.5*(lcdiff1+lcdiff2)
#Print the parameters and their default values:
for idefval in range(0,len(defVals.keys())):
thisdefval = defVals.keys()[idefval]
if thisdefval.find('_Im') > -1:
h('print "L5PC.apic[0].'+thisdefval+' = ", L5PC.apic[0].'+thisdefval+', "Default = ", '+str(defVals[thisdefval][1]))
else:
h('print "L5PC.soma[0].'+thisdefval+' = ", L5PC.soma[0].'+thisdefval+', "Default = ", '+str(defVals[thisdefval][0]))
# Determine if the variant obeys (isChanged = 0) or violates (isChanged = 1) the conditions I-V:
isChanged = nSpikes1 != 4 or nSpikes2 != 1 or nSpikes3 != 2 or spikfreqdiffrel > 0.1 or lcdiff > 600.0
print isChanged
if iter==0 and isChanged:
print "Even null mutation causes different spiking!! igene="+str(igene)+", imut="+str(imut)+", iallmutval="+str(iallmutval)
continue
if iter==1 and not isChanged:
print "This mutation effect does not alter spiking even when doubled!! igene="+str(igene)+", imut="+str(imut)+", iallmutval="+str(iallmutval)
continue
if iter>=2:
if isChanged:
nextCoeffs = [nextCoeffs[0],nextCoeffs[2],0.5*nextCoeffs[0]+0.5*nextCoeffs[2]]
else:
nextCoeffs = [nextCoeffs[2],nextCoeffs[1],0.5*nextCoeffs[1]+0.5*nextCoeffs[2]]
#Restore default values:
for imutvar in range(0,len(MT[igene][imut])):
mutvars = allmutvars[iallmutval][imutvar]
if type(mutvars) is str:
mutvars = [mutvars]
mutvals = allmutvals[iallmutval][imutvar]
for kmutvar in range(0,len(mutvars)):
newVal = defVals[mutvars[kmutvar]]
if mutvars[kmutvar].find('_Ih') > -1:
updateThese = [1,1,1]
elif mutvars[kmutvar].find('_Ca_HVA') > -1 or mutvars[kmutvar].find('_Ca_LVAst') > -1 or mutvars[kmutvar].find('_SKv3.1') > -1 or mutvars[kmutvar].find('_Ca_HVA') > -1 or mutvars[kmutvar].find('_SK_E2') > -1 or mutvars[kmutvar].find('_NaTa_t') > -1 or mutvars[kmutvar].find('_CaDynamics_E2') > -1:
updateThese = [1,1,0]
elif mutvars[kmutvar].find('_K_Pst') > -1 or mutvars[kmutvar].find('_K_Tst') > -1 or mutvars[kmutvar].find('_Nap_Et2') > -1:
updateThese = [1,0,0]
elif mutvars[kmutvar].find('_Im') > -1:
updateThese = [0,1,0]
else:
print "Error: str=" + str(mutvars[kmutvar])
updatedVars = [0,0,0]
for iupdated in range(0,3):
if updateThese[iupdated]:
h("""forsec L5PC."""+str(updatedVars[iupdated])+""" {
"""+mutvars[kmutvar]+""" = """+str(newVal[whichDefVal[iupdated]])+"""
}""")
theseCoeffsMut.append(nextCoeffs[0]+0.0)
# Save the results to be used by the drawfig1.py and others
picklelist = [nextCoeffs[0]+0.0,igene,imut,iallmutval,counter,MT]
file = open('scalings_cs'+str(icell)+'_'+str(counter)+'.sav', 'w')
pickle.dump(picklelist,file)
file.close()
theseCoeffsGene.append(theseCoeffsMut[:])
theseCoeffsAll.append(theseCoeffsGene[:])
theseCoeffsAllAll.append(theseCoeffsAll[:])
# After all scalings_csX_Y.sav have been saved, they should be collected to a single file by collectscalings.py