def parse_type(original):
'''parse a type string from the manifest allowing for some syntactic sugar'''
if not isinstance(original, str):
raise TypeError('Expected string but got %s instead (for: "%s")' %
(type(original).__name__, original))
value = original.lower()
alias_map = {
tt.LongType: ['int', 'long'],
tt.DoubleType: ['float', 'double'],
tt.BooleanType: ['bool', 'boolean'],
tt.StringType: ['str', 'string'],
}
for cl, aliases in alias_map.items():
if value in aliases:
return cl()
match = re.search(r'list\(([a-zA-Z0-9=;/\.\+]+)\)', original)
if match:
subtype = parse_type(match.group(1))
return tt.ListOf(subtype=subtype)
try:
get_mimetype(original)
return tt.URIType(category=original)
except ValueError:
pass
return object_from_string_call(original)
h5_bundle_mime_type = "application/unknown"
bundle = nestBundle2NeoH5_task.task.uri.build_bundle(h5_bundle_mime_type)
for gdf in input_files:
input_file = gdfio.GdfIO(gdf)
# in a bundle all neurons that spike at least once are registered
seg = input_file.read_segment(gdf_id_list=[],
t_start=t_start * pq.ms,
t_stop=t_stop * pq.ms)
output_filename = os.path.splitext(gdf)[0] + '.h5'
output_file = neo.io.NeoHdf5IO(output_filename)
output_file.write(seg.spiketrains)
output_file.close()
output_dst = os.path.basename(output_filename)
bundle.add_file(src_path=output_filename,
dst_path=output_dst,
bundle_path=output_dst,
mime_type=h5_bundle_mime_type)
return bundle.save("neo_h5_bundle")
if __name__ == '__main__':
input_filename = tt.URI('application/unknown',
'bundle_example/microcircuit_model_bundle')
t_start = 0.
t_stop = 300.
nestBundle2NeoH5_task(input_filename, t_start, t_stop)
res.append(sub_res)
# Close remaining file h5
ion.close()
print res
# Create plot
plt.title("LV Histogram")
plt.xlabel("LV")
plt.ylabel("")
# Plotting an histogram
plt.grid(True)
plt.hist(res, bins=100, normed=1, histtype='bar', rwidth=1)
# plt.show()
output_file = 'result_lv_histogram_task.png'
with open(output_file, 'w') as output:
plt.savefig(output)
return lv_histogram_task.task.uri.save_file(mime_type='image/png',
src_path=output_file,
dst_path=output_file)
if __name__ == '__main__':
# Run local test
filename = tt.URI('application/unknown', 'generate_poisson_spiketrains.h5')
lv_histogram_task(filename)
description: Neuron IDs in the input file that should be
extracted, e.g., [1, 2, 3]. Provide an empty list [] to
extract all neurons with at least one spike.
Returns:
res: application/unknown
"""
gdf = gdf2NeoH5_task.task.uri.get_file(gdf_file)
input_file = gdfio.GdfIO(gdf)
seg = input_file.read_segment(gdf_id_list=gdf_id_list,
t_start=t_start * pq.ms,
t_stop=t_stop * pq.ms)
output_filename = os.path.splitext(gdf)[0] + '.h5'
output_file = neo.io.NeoHdf5IO(output_filename)
output_file.write(seg.spiketrains)
output_file.close()
output_dst = output_filename.split('/')[-1]
return gdf2NeoH5_task.task.uri.save_file(mime_type='application/unknown',
src_path=output_filename,
dst_path=output_dst)
if __name__ == '__main__':
input_filename = tt.URI('application/unknown', 'spikes_L4E-77177-0.gdf')
t_start = 0.
t_stop = 300.
gdf_id_list = []
gdf2NeoH5_task(input_filename, t_start, t_stop, gdf_id_list)
for k in range(number_of_spike_trains):
spiketrains.append(ion.get("/" + "SpikeTrain_" + str(k)))
ion.close()
psth_as = time_histogram(spiketrains, binsize=binsize * pq.ms)
result = cubic.cubic(psth_as, alpha=alpha)
# Plot
plt.bar(np.arange(0.75, len(result[1]) + .25, 1), result[1], width=.5)
plt.axhline(alpha, ls='--', color='r')
plt.xlabel('$\\xi$')
plt.ylabel('P value')
plt.title('$\hat\\xi$=' + str(result[0]))
output_filename = os.path.splitext(h5_path)[0] + '.png'
with open(output_filename, 'w') as result_pth:
plt.savefig(result_pth, dpi=100)
dst_name = os.path.basename(output_filename)
return cubic_task.task.uri.save_file(mime_type='image/png',
src_path=output_filename,
dst_path=dst_name)
if __name__ == '__main__':
input_filename = tt.URI('application/unknown', 'spikes_L5I-44930-0.h5')
alpha = 0.05
binsize = 1
cubic_task(input_filename, binsize, alpha)
l_original = len(cc['original'])
dataset_orig = file.create_dataset("/cc_group/original",
(l_original, cc['original'][0].size),
dtype=h5py.h5t.NATIVE_FLOAT)
data = np.zeros((l_original, cc['original'][0].size))
for i in range(l_original):
for j in range(cc['original'][i].size):
data[i][j] = cc['original'][i].item(j)
dataset_orig[...] = data
if __name__ == '__main__':
# this number relates to the "-t" parameter:
# -t 0-X => number_of_jobs=X+1
# INPUT-second parameter
# number_of_jobs is (0, 200]
number_of_jobs = 1
# INPUT-third parameter
# job parameter: a number between 0 and number_of_jobs-1
import os
PBS_value = os.getenv('PBS_ARRAYID')
if PBS_value is not None:
job_id = int(PBS_value)
else:
job_id = 0
# INPUT-first parameter
inputdata = tt.URI('application/unknown', 'data/experiment.h5')
crosscorrelogram_task(inputdata, number_of_jobs, job_id)
from random import randint
from active_worker.task import task
from task_types import TaskTypes as tt
# This decoration is deprecated, from OLD-VERSION of task-sdk
_task_full_name = "elephant_cv_task"
_task_caption = "execute cv-function task"
_task_author = "Long Phan"
_task_description = "Input data is randomized inside function, used for\
calculate cv and output an image/png as result"
_task_categories = ['elephant']
_task_compatible_queues = ['cscs_viz']
@task(returns=tt.URIType('image/png'))
def elephant_cv_task():
# create data numpy for cv
data_list = []
for i in xrange(98):
range_size = randint(1, 10)
data_list.append(np.random.uniform(0, 1000, range_size))
# print data_list
# print len(data_list)
# res: list of results calculated by cv-function
res = []
# pass data as input-parameter to cv-function
for i in data_list:
res.append(cv(i))