def files_by_dist(files,dists,cid,endf):
#all_files,dists=get_cell_traces(cid,abfpath)
afd=[(file,dist) for file,dist in zip(files,dists)]
#afd.sort(key=lambda tup: tup[1])
fig=plt.figure(figsize=(20,20),frameon=False)
for filepath,distance in afd:
raw,block,segments,header=load_abf(filepath) #TODO fp segment dict like I had with queue/threading
if len(segments)==1:
continue
for segment in segments:
#for ass in segment.analogsignals:
ass=segment.analogsignals[0]
xs=np.linspace(0,.1,len(ass.times.base))+distance #center by distance
#ys=(ass.base-np.min(ass.base))/(np.max(ass.base)-np.min(ass.base))-segment.index*1.5 #move down by segment number
ys=(ass.base-np.mean(ass.base))/2.5+segment.index #move down by segment number
plt.plot(xs[:len(xs)//14]*1000,ys[:len(ys)//14],'k-',label='%s'%distance)
plt.ylim(-.2,5)
plt.yticks(np.arange(0,5,1))
plt.xlabel('Distance from cell body in $\mu$m. Normalized trial length')
plt.ylabel('Amplitude and run number')
plt.title('Example traces as a function of distance from cell %s'%cid)
#fig.frameon(False)
#fig.patch.set_visible(False)
ax=fig.axes[0]
ax.spines['left'].set_edgecolor((1,1,1,1))
format_axes(ax)
#embed()
fig.savefig('/tmp/fd_%s_%s.png'%(cid,endf[-4:]),bbox_inches='tight',pad_inches=0)
def plot_count_spikes(
filepath,
signal_map): #FIXME integrate this with count_spikes properly
""" useful for manual review """
raw, block, segments, header = load_abf(filepath)
if list(raw.read_header()['nADCSamplingSeq'][:2]) != [1, 2]: #FIXME
print('Not a ledstim file')
return None, None
nseg = len(segments)
gains = header['fTelegraphAdditGain'] #TODO
zero = transform_maker(0, 20 * gains[0]) #cell
one = transform_maker(0, 5) #led
plt.figure()
counts = []
for seg, n in zip(segments, range(nseg)):
if len(seg.analogsignals) != 2:
print('No led channel found.')
continue
plt.subplot(nseg, 1, n + 1)
nas = seg.size()['analogsignals']
signal = zero(seg.analogsignals[0])
led = one(seg.analogsignals[1])
led_on_index, base, maxV, minV = detect_led(
led) #FIXME move this to count_spikes
if not len(led_on_index):
print('No led detected maxV: %s minV: %s' % (maxV, minV))
continue
sig_on = signal.base[led_on_index]
sig_mean = np.mean(sig_on)
sig_std = np.std(sig_on)
#plt.plot(sig_std+sig_mean)
sm_arr = base * sig_mean
std_arr = base * sig_std
thresh = 3.3
s_list = detect_spikes(sig_on, thresh, 5)
counts.append(len(s_list))
[plt.plot(s, sig_on[s], 'ro') for s in s_list] #plot all the spikes
plt.plot(sig_on, 'k-')
plt.plot(sm_arr, 'b-')
plt.fill_between(np.arange(len(sm_arr)),
sm_arr - std_arr * thresh,
sm_arr + std_arr * thresh,
color=(.8, .8, 1))
plt.xlim((0, len(sig_on)))
plt.title('%s spikes %s' % (len(s_list), block.file_origin))
#plt.title(block.file_origin)
#plt.title('%s Segment %s'%(block.file_origin,n))
return np.mean(counts), np.var(counts), counts
def plot_abf(filepath,signal_map): #FIXME need better abstraction for the sigmap
raw,block,segments=load_abf(filepath)
nseg=len(segments)
plt.figure()
for seg,n in zip(segments,range(nseg)):
plt.title(block.file_origin)
#plt.title('%s Segment %s'%(block.file_origin,n))
nas=seg.size()['analogsignals']
for anasig,i in zip(seg.analogsignals,range(nas)):
plt.subplot(nas,1,i+1)
stp=signal_map[i](anasig)
plot_signal(stp)
def plot_abf(filepath,
signal_map): #FIXME need better abstraction for the sigmap
raw, block, segments = load_abf(filepath)
nseg = len(segments)
plt.figure()
for seg, n in zip(segments, range(nseg)):
plt.title(block.file_origin)
#plt.title('%s Segment %s'%(block.file_origin,n))
nas = seg.size()['analogsignals']
for anasig, i in zip(seg.analogsignals, range(nas)):
plt.subplot(nas, 1, i + 1)
stp = signal_map[i](anasig)
plot_signal(stp)
def plot_count_spikes(filepath,signal_map): #FIXME integrate this with count_spikes properly
""" useful for manual review """
raw,block,segments,header=load_abf(filepath)
if list(raw.read_header()['nADCSamplingSeq'][:2]) != [1,2]: #FIXME
print('Not a ledstim file')
return None,None
nseg=len(segments)
gains=header['fTelegraphAdditGain'] #TODO
zero=transform_maker(0,20*gains[0]) #cell
one=transform_maker(0,5) #led
plt.figure()
counts=[]
for seg,n in zip(segments,range(nseg)):
if len(seg.analogsignals) != 2:
print('No led channel found.')
continue
plt.subplot(nseg,1,n+1)
nas=seg.size()['analogsignals']
signal=zero(seg.analogsignals[0])
led=one(seg.analogsignals[1])
led_on_index,base,maxV,minV=detect_led(led) #FIXME move this to count_spikes
if not len(led_on_index):
print('No led detected maxV: %s minV: %s'%(maxV,minV))
continue
sig_on=signal.base[led_on_index]
sig_mean=np.mean(sig_on)
sig_std=np.std(sig_on)
#plt.plot(sig_std+sig_mean)
sm_arr=base*sig_mean
std_arr=base*sig_std
thresh=3.3
s_list=detect_spikes(sig_on,thresh,5)
counts.append(len(s_list))
[plt.plot(s,sig_on[s],'ro') for s in s_list] #plot all the spikes
plt.plot(sig_on,'k-')
plt.plot(sm_arr,'b-')
plt.fill_between(np.arange(len(sm_arr)),sm_arr-std_arr*thresh,sm_arr+std_arr*thresh,color=(.8,.8,1))
plt.xlim((0,len(sig_on)))
plt.title('%s spikes %s'%(len(s_list),block.file_origin))
#plt.title(block.file_origin)
#plt.title('%s Segment %s'%(block.file_origin,n))
return np.mean(counts),np.var(counts),counts
def count_spikes(filepath, signal_map): #TODO
raw, block, segments, header = load_abf(filepath)
nseg = len(segments)
gains = header['fTelegraphAdditGain'] #TODO
zero = transform_maker(0, 20 * gains[0]) #cell
one = transform_maker(0, 5) #led
counts = []
for seg, n in zip(segments, range(nseg)):
nas = seg.size()['analogsignals']
signal = zero(seg.analogsignals[0])
led = one(seg.analogsignals[1])
led_on_index = np.where(led.base < led.base[0] - .05)[0]
base = np.ones_like(led_on_index)
sig_on = signal.base[led_on_index]
thresh = 3.3
#print(filepath)
s_list = detect_spikes(sig_on, thresh, 5)
counts.append(len(s_list))
return np.mean(counts), np.var(counts), counts
def count_spikes(filepath,signal_map): #TODO
raw,block,segments,header=load_abf(filepath)
nseg=len(segments)
gains=header['fTelegraphAdditGain'] #TODO
zero=transform_maker(0,20*gains[0]) #cell
one=transform_maker(0,5) #led
counts=[]
for seg,n in zip(segments,range(nseg)):
nas=seg.size()['analogsignals']
signal=zero(seg.analogsignals[0])
led=one(seg.analogsignals[1])
led_on_index=np.where(led.base < led.base[0]-.05)[0]
base=np.ones_like(led_on_index)
sig_on=signal.base[led_on_index]
thresh=3.3
#print(filepath)
s_list=detect_spikes(sig_on,thresh,5)
counts.append(len(s_list))
return np.mean(counts),np.var(counts),counts
def files_by_dist(files, dists, cid, endf):
#all_files,dists=get_cell_traces(cid,abfpath)
afd = [(file, dist) for file, dist in zip(files, dists)]
#afd.sort(key=lambda tup: tup[1])
fig = plt.figure(figsize=(20, 20), frameon=False)
for filepath, distance in afd:
raw, block, segments, header = load_abf(
filepath) #TODO fp segment dict like I had with queue/threading
if len(segments) == 1:
continue
for segment in segments:
#for ass in segment.analogsignals:
ass = segment.analogsignals[0]
xs = np.linspace(0, .1, len(
ass.times.base)) + distance #center by distance
#ys=(ass.base-np.min(ass.base))/(np.max(ass.base)-np.min(ass.base))-segment.index*1.5 #move down by segment number
ys = (ass.base - np.mean(
ass.base)) / 2.5 + segment.index #move down by segment number
plt.plot(xs[:len(xs) // 14] * 1000,
ys[:len(ys) // 14],
'k-',
label='%s' % distance)
plt.ylim(-.2, 5)
plt.yticks(np.arange(0, 5, 1))
plt.xlabel('Distance from cell body in $\mu$m. Normalized trial length')
plt.ylabel('Amplitude and run number')
plt.title('Example traces as a function of distance from cell %s' % cid)
#fig.frameon(False)
#fig.patch.set_visible(False)
ax = fig.axes[0]
ax.spines['left'].set_edgecolor((1, 1, 1, 1))
format_axes(ax)
#embed()
fig.savefig('/tmp/fd_%s_%s.png' % (cid, endf[-4:]),
bbox_inches='tight',
pad_inches=0)
def dothing(args):
fp1, fp2, d1, d2 = args
fn = fp1.split('/')[-1]
figure = plt.figure(figsize=(20, 20))
spike_counts = []
spike_dist = None
base_counts = []
base_dist = None
for filepath, distance in zip((fp1, fp2), (d1, d2)):
raw, block, segments, header = load_abf(filepath)
if list(raw.read_header()['nADCSamplingSeq'][:2]) != [1, 2
]: #FIXME
print('Not a ledstim file')
return None, None
#print(header['nADCSamplingSeq']) #gain debugging
#print(header['nTelegraphEnable'])
#print(header['fTelegraphAdditGain'])
#gains=header['fTelegraphAdditGain'] #TODO
#print(fn,gains)
#TODO cell.headstage number?!
#headstage_number=1 #FIXME
#gain=gains[headstage_number] #FIXME the problem I think is still in AxonIO
#zero=transform_maker(0,20*gain) #FIXME where does the first 20 come from !?
#one=transform_maker(0,5) #led
#one=transform_maker(0,10) #led no idea why the gain is different for these
nseg = len(segments)
for seg, n in zip(segments, range(nseg)):
title_base = ''
if nseg == 1:
plt.subplot(6, 1, 6)
base_fp = filepath
else:
dist_fp = filepath
plt.subplot(6, 1, n + 1)
signal = seg.analogsignals[0].base
units_sig = seg.analogsignals[0].units
sig_times = seg.analogsignals[0].times.base
units_sig_times = seg.analogsignals[0].times.units
led = seg.analogsignals[1].base
led_on_index, base, maxV, minV = detect_led(
led) #FIXME move this to count_spikes
if not len(led_on_index):
print('No led detected maxV: %s minV: %s' % (maxV, minV))
continue
sig_on = signal[led_on_index]
sig_on_times = sig_times[
led_on_index] #FIXME may not be synch?
sig_mean = np.mean(sig_on)
sig_std = np.std(sig_on)
#plt.plot(sig_std+sig_mean)
sm_arr = base * sig_mean
sm_arr_times = sig_on_times #[led_on_index]
std_arr = base * sig_std
#do spike detection
if spikes:
spike_indexes, threshold = detect_spikes(
sig_on, std_thrs, std_max, threshold_func(filepath))
spike_times = sig_on_times[spike_indexes]
if spike_func(filepath):
sc = spike_func(filepath)[seg.index]
else:
sc = len(spike_indexes)
if nseg == 1:
base_counts.append(sc)
base_dist = distance
else:
spike_counts.append(sc)
spike_dist = distance
#TODO find a way to associate the spikecount with the DataFile
plt.plot(spike_times, sig_on[spike_indexes], 'ro')
title_base += '%s spikes ' % (sc)
if do_plot:
title_base += '%s $\mu$m from cell %s $\\verb|%s|$ ' % (
int(distance * 1000), cell_id, block.file_origin[:-4])
#title_base+='gain = %s *20'%gain
plt.title(title_base)
plt.xlabel(units_sig_times)
plt.ylabel(units_sig)
#plt.xlabel(led.times.units)
#plt.ylabel(led.units)
#plt.plot(led.times.base[led_on_index],led.base[led_on_index])
lrf = (sig_on_times[0], sig_on_times[-1])
plt.plot(sig_on_times[::10],
sig_on[::10],
'k-',
linewidth=1)
plt.plot(lrf, [sig_mean] * 2, 'b-', label='sig mean')
plt.plot(lrf, [threshold] * 2, 'm-', label='threshold')
plt.plot(lrf, [sig_mean + sig_std * std_max] * 2,
'y-',
label='max thresh')
if threshold_func(filepath):
plt.plot(lrf, [threshold_func(filepath)] * 2,
'g-',
label='manual thresh')
plt.fill_between(np.arange(len(sm_arr)),
sm_arr - std_arr * std_thrs,
sm_arr + std_arr * std_thrs,
color=(.8, .8, 1))
plt.xlim((sig_on_times[0], sig_on_times[-1]))
plt.legend(loc='upper right',
fontsize='xx-small',
frameon=False)
spike_count_callback(base_counts, spike_counts, base_dist, spike_dist,
dist_fp, base_fp)
spath = '/tmp/' + str(cell_id) + '_' + fn[:-4] + '.png'
if do_plot:
print(spath)
plt.savefig(spath, bbox_inches='tight', pad_inches=0)
#plt.show()
figure.clf() #might reduce mem footprint
plt.close()
gc.collect()
def dothing(args):
fp1,fp2,d1,d2=args
fn=fp1.split('/')[-1]
figure=plt.figure(figsize=(20,20))
spike_counts=[]
spike_dist=None
base_counts=[]
base_dist=None
for filepath,distance in zip((fp1,fp2),(d1,d2)):
raw,block,segments,header=load_abf(filepath)
if list(raw.read_header()['nADCSamplingSeq'][:2]) != [1,2]: #FIXME
print('Not a ledstim file')
return None,None
#print(header['nADCSamplingSeq']) #gain debugging
#print(header['nTelegraphEnable'])
#print(header['fTelegraphAdditGain'])
#gains=header['fTelegraphAdditGain'] #TODO
#print(fn,gains)
#TODO cell.headstage number?!
#headstage_number=1 #FIXME
#gain=gains[headstage_number] #FIXME the problem I think is still in AxonIO
#zero=transform_maker(0,20*gain) #FIXME where does the first 20 come from !?
#one=transform_maker(0,5) #led
#one=transform_maker(0,10) #led no idea why the gain is different for these
nseg=len(segments)
for seg,n in zip(segments,range(nseg)):
title_base=''
if nseg == 1:
plt.subplot(6,1,6)
base_fp=filepath
else:
dist_fp=filepath
plt.subplot(6,1,n+1)
signal=seg.analogsignals[0].base
units_sig=seg.analogsignals[0].units
sig_times=seg.analogsignals[0].times.base
units_sig_times=seg.analogsignals[0].times.units
led=seg.analogsignals[1].base
led_on_index,base,maxV,minV=detect_led(led) #FIXME move this to count_spikes
if not len(led_on_index):
print('No led detected maxV: %s minV: %s'%(maxV,minV))
continue
sig_on=signal[led_on_index]
sig_on_times=sig_times[led_on_index] #FIXME may not be synch?
sig_mean=np.mean(sig_on)
sig_std=np.std(sig_on)
#plt.plot(sig_std+sig_mean)
sm_arr=base*sig_mean
sm_arr_times=sig_on_times #[led_on_index]
std_arr=base*sig_std
#do spike detection
if spikes:
spike_indexes,threshold=detect_spikes(sig_on,std_thrs,std_max,threshold_func(filepath))
spike_times=sig_on_times[spike_indexes]
if spike_func(filepath):
sc=spike_func(filepath)[seg.index]
else:
sc=len(spike_indexes)
if nseg == 1:
base_counts.append(sc)
base_dist=distance
else:
spike_counts.append(sc)
spike_dist=distance
#TODO find a way to associate the spikecount with the DataFile
plt.plot(spike_times,sig_on[spike_indexes],'ro')
title_base+='%s spikes '%(sc)
if do_plot:
title_base+='%s $\mu$m from cell %s $\\verb|%s|$ '%(int(distance*1000),cell_id,block.file_origin[:-4])
#title_base+='gain = %s *20'%gain
plt.title(title_base)
plt.xlabel(units_sig_times)
plt.ylabel(units_sig)
#plt.xlabel(led.times.units)
#plt.ylabel(led.units)
#plt.plot(led.times.base[led_on_index],led.base[led_on_index])
lrf=(sig_on_times[0],sig_on_times[-1])
plt.plot(sig_on_times[::10],sig_on[::10],'k-',linewidth=1)
plt.plot(lrf,[sig_mean]*2,'b-', label = 'sig mean' )
plt.plot(lrf,[threshold]*2,'m-', label = 'threshold' )
plt.plot(lrf,[sig_mean+sig_std*std_max]*2,'y-', label = 'max thresh' )
if threshold_func(filepath):
plt.plot(lrf,[threshold_func(filepath)]*2,'g-', label = 'manual thresh' )
plt.fill_between(np.arange(len(sm_arr)),sm_arr-std_arr*std_thrs,sm_arr+std_arr*std_thrs,color=(.8,.8,1))
plt.xlim((sig_on_times[0],sig_on_times[-1]))
plt.legend(loc='upper right',fontsize='xx-small',frameon=False)
spike_count_callback(base_counts,spike_counts,base_dist,spike_dist,dist_fp,base_fp)
spath='/tmp/'+str(cell_id)+'_'+fn[:-4]+'.png'
if do_plot:
print(spath)
plt.savefig(spath,bbox_inches='tight',pad_inches=0)
#plt.show()
figure.clf() #might reduce mem footprint
plt.close()
gc.collect()