def run(self, event=None): i=self.chooseMode1.GetSelection() sdf=self.chooseMode1.GetString(i) sdf=evtdet.DETECTORS[sdf] pars=self.discPars1.GetValue() pars=eval(pars) if type(pars) in [float, int]: pars=(pars,) else: pars=tuple(pars) sel=(None, [self.preferences["Trigger Channel"]], None) path=self.preferences["Dpath"]+'/spikes1' if self.dat.getSubData(path): self.dat.getSubData(path).sever() sdf(self.dat, sel, pars, path) eventCondition(self.dat, dpath=path, select=sel, lead=self.preferences["Template Lead (ms)"], length=self.preferences["Template Length"], newpath=self.preferences["Dpath"]+'/evts1', milliseconds=True) sel=(None, [self.preferences["Output Channel"]], None) eventCondition(self.dat, dpath=path, select=sel, lead=self.preferences["Template Lead (ms)"], length=self.preferences["Template Length"], newpath=self.preferences["Dpath"]+'/evts2', milliseconds=True) self.showEvts(self.graph1, self.preferences["Dpath"]+'/evts1') self.showPV(self.graph1, 0, self.preferences["Dpath"]+'/evts1') self.showEvts(self.graph2, self.preferences["Dpath"]+'/evts2') self.showPV(self.graph2, 1, self.preferences["Dpath"]+'/evts2')
def conditionEach(ds, dpathFrom='/', dpathEventsToConditionWith='/evts', dpathStimulus='/avstims'):
'''
Applies gicspikesort.conditioning.eventCondition to all timeseries and events in ds
'''
dod = ds.getHierarchy()
numberNames=dod[dpathFrom].getData()[:,1]
for k in range(max(numberNames)+1):
#print(k)
#separate each unit to its own channel
mm = list(numberNames).index(k)
tmpName = '/'+ds.getLabels()[k]
hdr = {'SampleType':'events','SamplesPerSecond':5e4, 'Labels':ds.getLabels()[k]}
dta = dod[dpathFrom].getData()[mm:mm+list(numberNames).count(k),0]
ds.createSubData(tmpName,dta, head=hdr)
#event condition
newName='/sorted_'+ds.getLabels()[k]
gicc.eventCondition(ds, dpathEventsToConditionWith, tmpName, 5, 35, newName, False)
#remove the new channel
dod = ds.getHierarchy()
dod[tmpName].sever()
def tmpWork(ds, prefChans=None):
'''
Just the day's workflow, save keystrokes
'''
dataDict = ds.getHierarchy()
#set prefChans
if not prefChans:
prefChans = range(ds.shape()[1]-4,ds.shape()[1])
print('Selecting stimulus channels')
#print(prefChans)
#get a marker of the base channel starting each stimulus seti
_getStims(ds,prefChans, .006,.003)
#remove the last element from isolateSpikes WIP
#events=dataDict['/tmpname'].getData()
#events=events[0:len(events)/2-1,:]
#dataDict['/tmpname'].setData(events)
#print(dataDict['/tmpname'].shape())
#trigger on these marks to get the average stimulus
gicc.eventCondition(ds,'/evts', '/', 40.0, 80.0, '/avstim', milliseconds=True)
#remove the channel with multiple marks per stimulus run
dataDict = ds.getHierarchy()
repNumber=dataDict['/avstim'].header()['Reps']
#print(repNumber)
#average just the spikes and the base and tip stim
average = dataDict['/avstim'].getData()
newdata = average[:,0:repNumber-1].mean(1)
johnsData=average[:,0:repNumber-1]
#try to find the base and stimulus channels, or arbitrarily set base and stim channels
labelNames = ds.getLabels()
if 'nozzle 1 (No Comment)' in labelNames:
basenum = labelNames.index('nozzle 1 (No Comment)')
tipnum = labelNames.index('nozzle 2 (No Comment)')
else:
basenum = prefChans[m]
tipnum = prefChans[1]
for m in range(basenum,basenum+4):
nd2=average[:,repNumber*m:repNumber*(m+1)-1].mean(1)
newdata=n.column_stack((newdata,nd2))
#nd2=average[:,repNumber*basenum:repNumber*(basenum+1)-1].mean(1)
#newdata=n.column_stack((newdata,nd2))
#nd2=average[:,repNumber*(basenum+1):repNumber*(basenum+2)-1].mean(1)
#newdata=n.column_stack((newdata,nd2))
hdr = {'SampleType':'timseries','SamplesPerSecond':ds.header()['SamplesPerSecond']}
ds.datinit(newdata,hdr)
dataDict['/avstim'].sever()
ds.createSubData('/johns',johnsData, head=hdr)
#dataDict['/'].sever()
#remove the non stimulus channels from the conditioned data
#evtsData = dataDict['/evts'].getData()
#evtsNumber = len(evtsData)
#evtsNumber=evtsNumber*10
#resizeData=dataDict['/avstim'].getData()[:,evtsNumber:]
#dataDict['/avstim'].sever()
#hdr = dataDict['/avstim'].header()
#ds.createSubData('/avstim',resizeData,hdr)
#save the new file
newFName = ds.xpath(True)[0].fileinformation.get('filename','NoNameGiven')
newFName = newFName[0:newFName.rfind('.')] + '.mat'
IO.write(ds,newFName,newdoc=True)
def dividebyStims(ds, prefChans=[9,10,11,12], thresh1=.5, thresh2=.3, path=None, stimchantag=None, newbasename=''):
'''
Find all the stimuli and sort each run into a limited number of classes, then put each class together in a new file
Assumptions: 1) On repeated stimuli we're going faster than 30Hz
2) We have at least 200ms between stimuli repeats
3) Recording doesn't start in the middle of a stimulus
4) Base is the first stimulus channel, tip is second
'''
#stimoptions=array([-9.,-7.,-5.,-3.,-1.,0.,1.,3.,5.,7.,9.])
#stimoptions=array([-60,-55,-50,-45,-40,40,45,50,55,60])
#stimoptions=array([-115,-100,-85,-60,-45,-30,-15,15,30,45,60,85,100,115])
winl = 100*ds.header()['SamplesPerSecond']/1e3
winh = 200*ds.header()['SamplesPerSecond']/1e3
if stimchantag:
jnk, prefChans = _stims_from_strng(ds,stimchantag)
else:
_findStims(ds,prefChans,thresh1,thresh2,pth=path)
dataDict = ds.getHierarchy()
tips = dataDict['/tmpname'].getData()
#print tips, type(tips), shape(tips)
#print len(tips), tips
cheatvar = 0
if not len(tips): #without stimuli detected on the channels in question we have some trouble with the following function. Which wouldn't be a problem except THAT SOMEWHERE THIS FUNCTION RESIZES DATA, AND WE NEED TO RESIZE IN THE ABSENCE OF STIMULI AS WELL. BUT WHILE ITS POSSIBLE TO RUN THROUGH THE PROGRAM, IT SOMEHOW STILL MISSES THE PLACE WHERE THINGS ARE RESIZED`
tips = array([[0.,0.],[0.,1.],[0.,2.],[0.,3.],[5.,0.],[5.,1.],[5.,2.],[5.,3.]])
cheatvar = 1
ds.getSubData('/tmpname').sever()
ds.createSubData('/tmpname',data=tips,head={})
return True
#print tips, type(tips), shape(tips)
# if newbasename == '':
# newfname = ds.xpath(True)[0].fileinformation.get('filename','nonamegiven')
# else:
# newfname = newbasename
# newfname = os.path.splitext(newfname)[0] + 'pt1.mien'
# print 'No stimuli found, saving only {0}.'.format(newfname)
# save_wo_leak(ds, newfname)
# return
#take care of uneven numbers of stims on various channels
runlengths = [len(tips[tips[:,1]==m,0]) for m in list(set(tips[:,1]))]
shortchan = argmin(runlengths)
others = list(set(range(len(runlengths)))-set([shortchan]))
longchan=others[0]
if not runlengths[shortchan]==runlengths[longchan]:
frontdif = abs(tips[tips[:,1]==shortchan,0][0] - tips[tips[:,1]==longchan,0][0])
endif = abs(tips[tips[:,1]==shortchan,0][-1] - tips[tips[:,1]==longchan,0][-1])
badend =int(where(frontdif>endif,0,-1))
tt=list(tips)
for m in range(len(others)-1,-1,-1):
tt.pop(where(tips[:,1]==others[m])[0][badend])
tips = array(tt)
stims = tips[where(tips[:,1]==0)[0],0]
print 'bb', stims, range(1,len(prefChans))
for m in range(1,len(prefChans)):
addons = tips[where(tips[:,1]==m)[0],0]
if not addons.shape[0] in stims.shape: #so far only fixed when one of addons not equal to number 2 is the wrong size
if addons.shape[0]>max(stims.shape):
raise NameError('do not know how to do')
else:
while addons.shape[0] != max(stims.shape):
if addons[0]-stims[0,0] > addons[-1]-stims[0,-1]: #the double part reference to stims eliminates the chance to catch addons number 2 wrong size
addons = hstack((0,addons))
else:
addons = hstack((addons,0))
#return stims, addons
stims = vstack((stims, addons))
stims = transpose(stims)
#print 'aa', stims
#datadict = ds.gethierarchy()
#tips = datadict['/tmpname1'].getdata()
#bases = datadict['/tmpname0'].getdata()
#ch3 = datadict['/tmpname2'].getdata()
#tips2 = datadict['/tmpname3'].getdata()
#stims = n.hstack((bases,tips,ch3,tips2))
stims = _reformatted_stims(stims, ds.header()['SamplesPerSecond'])
#print 'vv', shape(stims)
# stimoptions = _unique_stims(stims)
# stimoptions, newstim = _translate_stims(stimoptions)
#print 'hello', stims, stims.shape, type(stims), '\n\n\n', fubaru
stims, newstim = _translate_stims(stims)
#print 'cc', shape(stims), newstim
# stimoptions=array(stimoptions)
events = _unique_events_from_stims(ds)
if min(events)[0] < 30/1e3*ds.header()['SamplesPerSecond']:
print('this recording may be starting in the middle of a stimulus and subsequent analysis will get wonky:- sep.dsp.dividebystims')
stimlist = list(stims)
if cheatvar:
stimlist = [0]
stimoptions=array(list(set(stimlist)))
#clean up after myself
dataDict['/tmpname'].sever()
#datadict['/tmpnametot'].sever()
#datadict['/tmpname0'].sever()
#datadict['/tmpname1'].sever()
#datadict['/tmpname2'].sever()
#datadict['/tmpname3'].sever()
#datadict['/evts'].sever()
#see if the spikes are already identified in this file and need to be distributed works ok without sorted spikes< but need some way to get "/hidden/stimulus' copied as well
tempsin=ds.getElements()
tempsin=[t for t in tempsin if t.name().startswith("spikesort_")]
nams=[]
temps=[]
for m in range(len(tempsin)):#for each unit...
temps.append(tempsin[m].getElements(attribs={'SampleType' :'events'},depth=1)[0])
nams.append(tempsin[m].name())
temps[m]=temps[m].getData()#get the event times off all spikes for a given unit
#pull out chunks for each repeat, paste them together, save new file
ds2=ds.clone(False)
channum=ds.shape()[1]
hdr = ds.header()
#hd2 = {'SamplesPerSecond':50000.0,'StartTime':0.0,'SampleType':'events'}
for m in range(len(stimoptions)):
#print 'right here', stimoptions, m, events, stimlist, array(stimlist), stimoptions[m], len(events[array(stimlist)==stimoptions[m]])
togets=events[array(stimlist)==stimoptions[m]]#get a list of all stimuli matching this stimoption
data = n.zeros((1,channum))#preallocate
locevts = []#preallocate
for r in range(len(tempsin)):
locevts.append([])#preallocate
for r in range(len(togets)):#for each matching stimulus event
data=n.concatenate((data,dataDict['/'].getData()[int(togets[r]-winl):int(togets[r]+winh),:]))#append the chunk winl before the stimulus to winh after the stimulus to data
for q in range(len(temps)):#for each unit...
under=set(temps[q][temps[q]>togets[r]-winl])# find all the 'temps' in the appropriate within winl to winh of the event
over=set(temps[q][temps[q]<togets[r]+winh])
ok=array(list(under.intersection(over)))
for t in range(len(ok)):#for each individual spike of a certian unit...
locevts[q].append(ok[t]-int(togets[r])+r*(winl+winh)+winl)#add each approved spike with the appropriate shift for how far along in the recording we are
ds2.datinit(data,hdr)
ds2.setAttrib('StartTime',0)
for q in range(len(temps)):
hd2['Labels']=nams[q]
ds2.createSubData('/'+nams[q],data=locevts[q],head=hd2)
if newbasename == '':
newfname = ds.xpath(True)[0].fileinformation.get('filename','nonamegiven')
else:
newfname = newbasename
if newfname==None:
print 'could not read filename'
newfname = os.path.expanduser('~') + '/Desktop/tempfile.mien'
newfname = os.path.splitext(newfname)[0] + 'pt' + str(m+1) +'.mien'
#print newfname
save_wo_leak(ds2,newfname)
if len(temps): #put this in standard format if the sorted spikes are already there
ds2.createSubData('/evts',data=array(range(len(togets)))*(winl+winh)+winl,head=hd2)
#get the data I want
gicc.eventCondition(ds2,'/evts', '/', 120.0, 160.0, '/avstim', milliseconds=True)
dataDict2 = ds2.getHierarchy()
repNumber = dataDict2['/avstim'].header()['Reps']
average = dataDict2['/avstim'].getData()
newdata = average[:,0:repNumber-1].mean(1)
#also get raw traces of the full recording
fulldata = average[:,0:repNumber-1]
for q in range(1,dataDict2['/'].shape()[1]):
fulldata = n.column_stack((fulldata,average[:,repNumber*q:repNumber*(q+1)-1]))
#get the stimulus channels I want
ms2pts=ds.header()['SamplesPerSecond']*1e-3
for q in range(len(prefChans)):
upper=int(togets[0]-120*ms2pts)
lower=int(togets[0]+40*ms2pts)
nd2 = getSelection(ds,(path,prefChans[q],None))[upper:lower]
newdata=n.column_stack((newdata,nd2))
#write over ds2 so it doesn't have needless data as well
avstim=dataDict2['/avstim'].getData()
hd3=dataDict2['/avstim'].header()
ds2.clearAll()
hdr = {'SampleType':'timseries','SamplesPerSecond':ds.header()['SamplesPerSecond']}
ds2.datinit(newdata,hdr)
ds2.createSubData('/evts',data=array(range(len(togets)))*(winl+winh)+winl,head=hd2)
ds2.createSubData('/avstim',avstim,head=hd3)
ds2.createSubData('/fullstim',fulldata,head=hd3)
#add the spikes in a single piece of data
locdata=zeros((2,1))
for q in range(len(temps)):
indata=vstack((locevts[q],q*ones((1,len(locevts[q])))))
locdata=hstack((locdata,indata))
locdata=locdata[:,1:]
hd3=hd2.copy()
hd3['SampleType']='labeledevents'
hd3['Labels']=nams
ds2.createSubData('/spks',locdata.transpose(),head=hd3)
nFN2=newFName[:-5]+'spks.mat'
#print nFN2
save_wo_leak(ds2,nFN2)
ds2.clearAll()
ds2.sever()
ds.sever()
