def test_fetch_adjacencies(client):
bodies = [
294792184, 329566174, 329599710, 417199910, 420274150, 424379864,
425790257, 451982486, 480927537, 481268653
]
neuron_df, roi_conn_df = fetch_adjacencies(NC(bodyId=bodies),
NC(bodyId=bodies))
# Should not include non-primary ROIs (except 'NotPrimary')
assert not ({*roi_conn_df['roi'].unique()} - {*fetch_primary_rois()} -
{'NotPrimary'})
#
# For backwards compatibility with the previous API,
# You can also pass a list of bodyIds to this function (instead of NeuronCriteria).
#
bodies = [
294792184, 329566174, 329599710, 417199910, 420274150, 424379864,
425790257, 451982486, 480927537, 481268653
]
neuron_df2, roi_conn_df2 = fetch_adjacencies(bodies, bodies)
# Should not include non-primary ROIs (except 'NotPrimary')
assert not ({*roi_conn_df2['roi'].unique()} - {*fetch_primary_rois()} -
{'NotPrimary'})
assert (neuron_df.fillna('') == neuron_df2.fillna('')).all().all()
assert (roi_conn_df == roi_conn_df2).all().all()
# What happens if results are empty
neuron_df, roi_conn_df = fetch_adjacencies(879442155, 5813027103)
assert len(neuron_df) == 0
assert len(roi_conn_df) == 0
assert neuron_df.columns.tolist() == ['bodyId', 'instance', 'type']
def test_fetch_adjacencies(client):
bodies = [
294792184, 329566174, 329599710, 417199910, 420274150, 424379864,
425790257, 451982486, 480927537, 481268653
]
neuron_df, roi_conn_df = fetch_adjacencies(SC(bodyId=bodies),
SC(bodyId=bodies))
# Should not include non-primary ROIs (except 'NotPrimary')
assert not ({*roi_conn_df['roi'].unique()} - {*fetch_primary_rois()} -
{'NotPrimary'})
#
# For backwards compatibility with the previous API,
# You can also pass a list of bodyIds to this function (instead of SegmentCriteria).
#
bodies = [
294792184, 329566174, 329599710, 417199910, 420274150, 424379864,
425790257, 451982486, 480927537, 481268653
]
neuron_df2, roi_conn_df2 = fetch_adjacencies(bodies, bodies)
# Should not include non-primary ROIs (except 'NotPrimary')
assert not ({*roi_conn_df2['roi'].unique()} - {*fetch_primary_rois()} -
{'NotPrimary'})
assert (neuron_df.fillna('') == neuron_df2.fillna('')).all().all()
assert (roi_conn_df == roi_conn_df2).all().all()
def main():
parser = argparse.ArgumentParser()
parser.add_argument('--neuprint-server', '-n', default='neuprint.janelia.org')
parser.add_argument('--dataset', '-d')
parser.add_argument('--init', '-i', choices=['groundtruth', 'random'])
parser.add_argument('--verbose', '-v', action='store_true')
parser.add_argument('--debug', action='store_true')
parser.add_argument('--min-weight', '-w', default=10, type=int)
args = parser.parse_args()
c = Client(args.neuprint_server, args.dataset)
export_dir = f"{c.dataset}-w{args.min_weight}-from-{args.init}"
os.makedirs(export_dir, exist_ok=True)
# Fetch connectome (and export)
with Timer("Fetching/exporting connectome", logger):
criteria = NC(status='Traced', cropped=False, client=c)
neuron_df, roi_conn_df = fetch_adjacencies(criteria, criteria, min_total_weight=args.min_weight, export_dir=export_dir, properties=['type', 'instance'], client=c)
conn_df = roi_conn_df.groupby(['bodyId_pre', 'bodyId_post'], as_index=False)['weight'].sum()
strong_connections_df, g, nbs, partition_df = infer_hierarchy(neuron_df,
conn_df,
args.min_weight,
args.init,
args.verbose,
args.debug)
with Timer("Exporting inference results", logger):
pickle.dump(g, open(f'{export_dir}/graph.pkl', 'wb'))
pickle.dump(nbs, open(f'{export_dir}/nested-block-state.pkl', 'wb'))
pickle.dump(partition_df, open(f'{export_dir}/partition_df.pkl', 'wb'))
pickle.dump(strong_connections_df, open(f'{export_dir}/strong_connections_df.pkl', 'wb'))
logger.info("DONE")
def pn_kc_connections(properties=None,
sum_across_rois=False,
checks=True,
**kwargs):
"""
Returns `neuron_df`, `connection_df` as `neuprint.fetch_adjacencies`, but
only for PN->KC connections.
See also keyword arguments added by the `@fetch_function` decorator.
Keywords not covered above are passed to `neuprint.fetch_adjacencies`.
"""
if properties is None:
# TODO TODO TODO replace hardcoded properties w/ something enumerated
# using a cypher query
properties = nc
# TODO TODO TODO the docs make this sound like it might only return weights
# for connections shared by ALL neurons matched. is this true? see also the
# two kwargs dealing with minimum weights
# TODO consider using rois=[<appropriate str for calyx>] to just match
# those. compare results to those from manually filtering output not
# specifying those rois.
neuron_df, conn_df = nu.fetch_adjacencies(
# TODO if i end up factoring some of this fn into something to be shared
# across fns that get connections from one type to another, maybe just
# test whether input strs have asterisk in them, and set regex=True if
# so? first check if * is in any of the types that exist in the db!!!
NC(type='.*PN.*', regex=True),
NC(type='.*KC.*', regex=True),
# Default is just ['type','instance']
properties=properties,
**kwargs)
# TODO test the number of things returned above is the same as when doing
# the equivalent CONTAINS query in the web interface
# (also check just .* as suffix, not also prefix)
if checks:
# There CAN be (pre ID, post ID) duplicates, because weights are
# reported per ROI, so need to sum across ROIs if we just want
# the total weight between two neurons, ignoring where the synapses are.
assert not conn_df.duplicated(
[c for c in conn_df.columns if c != 'weight']).any()
# TODO TODO technically only do this check if include_nonprimary=False
# (so check kwargs) (if it's True, would need to do another check)
assert (set(conn_df.roi.unique()) -
set(nu.fetch_primary_rois()) == {'NotPrimary'})
if sum_across_rois:
conn_df = conn_df.groupby(['bodyId_pre',
'bodyId_post']).weight.sum().reset_index()
# No need to call filter_nontraced_or_cropped, at least as of 2020-05-31,
# where it had no effect.
return neuron_df, conn_df
def fetch_adjacency(criteria=None,
prefix='noncropped_traced',
force_download=False,
neuprint=conf.enable_neuprint,
adjpath=None,
**kwargs):
'''
simple neuprint.fetch_adjacencies wrapper.
Checks whether datasets were already downloaded and loads them accordingly.
By default func loads traced and noncropped neurons.
'''
assert not force_download or neuprint, 'no neuprint; cannot download dataset'
#compose adjpath
if adjpath is None:
if not isinstance(kwargs['rois'], list):
kwargs['rois'] = [kwargs['rois']]
datadir = conf.datasets_dir
postfix = '_' + '.'.join(
kwargs['rois']) if 'rois' in kwargs.keys() else ''
adjpath = os.path.join(datadir, prefix + postfix)
roipath = os.path.join(adjpath, conf.roi_connections_file)
neurpath = os.path.join(adjpath, conf.neurons_file)
files_exist = os.path.exists(adjpath) and os.path.exists(
roipath) and os.path.exists(neurpath)
assert files_exist or neuprint, 'no neuprint; cannot find dataset (and no way of fetching)'
if neuprint:
if 'client' not in kwargs.keys():
kwargs['client'] = Client(conf.neuprint_URL, conf.dataset_version,
conf.api_token)
if criteria == None:
criteria = NeuronCriteria(status='Traced',
cropped=False,
client=kwargs['client'])
print('dataset in adjpath=', adjpath)
if os.path.exists(adjpath) and os.path.exists(roipath) and os.path.exists(
neurpath) and not force_download:
print('dataset already downloaded')
adj = pd.read_csv(roipath)
neurons = pd.read_csv(neurpath)
else:
print('downloading dataset')
print(criteria)
neurons, adj = fetch_adjacencies(sources=criteria,
targets=criteria,
export_dir=adjpath,
**kwargs)
return neurons, adj
def get_adjacencies_neuprint(self,
accessions,
threshold=1,
testmode=False): # add testmode?
#fetch neuron-neuron connectivity for only between the accessions and only within PRIMARY rois and collapse rois to total
neuron_df, conn_df = fetch_adjacencies(sources=accessions,
targets=accessions)
conn_df = conn_df.groupby(['bodyId_pre', 'bodyId_post'],
as_index=False)['weight'].sum()
conn_df.rename(columns={
'bodyId_pre': 'source',
'bodyId_post': 'target'
},
inplace=True)
#filter by synapse count threshold (using total neuron-neuron connectivity in whole brain
conn_df = conn_df[conn_df.weight > threshold]
return conn_df
def build_matrix(results):
ntnstatus('Downloading structural information from Janelia')
bids = list(results['bodyId'].values) # Get bids of neurons to consider
neuron_df, conn_df = fetch_adjacencies(
bids, bids) # Get all synapses between neurons, and their weight
conn_df2 = conn_df.groupby(
['bodyId_pre', 'bodyId_post'], as_index=False)['weight'].sum(
) # Combine repeated edges and add their weights
#Dictionary of bid to vertex id
bodyIdtoId = {bids[i]: i for i in range(len(bids))}
#Save results in coo format
row = [bodyIdtoId[i] for i in conn_df2['bodyId_pre'].values]
col = [bodyIdtoId[i] for i in conn_df2['bodyId_post'].values]
data = [i for i in conn_df2['weight'].values]
N = len(bids)
mcoo = sparse.coo_matrix((data, (row, col)), shape=(N, N))
np.save(root + "structure/drosophila_weighted.npy", mcoo)
np.save(root + "structure/drosophila_bid2vertex.npy", bodyIdtoId)
import sys
sys.path.append('..')
import fly
os.makedirs('../data', exist_ok=True)
os.makedirs('../processed', exist_ok=True)
roi = sys.argv[1]
print('Downloading', roi)
print()
criteria = NeuronCriteria(status="Traced", cropped=False)
fetch_adjacencies(criteria, criteria, export_dir='../data/noncropped_traced_{}'.format(roi), rois=[roi])
df = pd.read_csv('../data/noncropped_traced_{}/roi-connections.csv'.format(roi))
dfroi = df[df.roi==roi]
neurons = sorted(list(set(dfroi.bodyId_post.unique()).union(set(dfroi.bodyId_pre.unique()))))
N = len(neurons)
neuron_ids = np.zeros(N, dtype=int)
neuron = dict()
for i, nid in enumerate(neurons):
neuron[nid] = i
neuron_ids[i] = nid
def fetch_adj_and_merge(criteria):
default_criteria_for_test = NC(inputRois='AB(L)')
neuron_df, conn_df = fetch_adjacencies(criteria, default_criteria_for_test)
conn_df = merge_neuron_properties(neuron_df, conn_df, ['type', 'instance'])
return conn_df
]])
print(roi_counts_df.query('bodyId==5813128308')) # synapse counts for ROIs
# print(neuron_df.columns)
# neuron_df, roi_counts_df = fetch_neurons(NC(type='MBON.*', regex=True)) # get mushroom body output neurons
# ------------------------------------------
# Fetch connections
# ------------------------------------------
''' Find synaptic connection strengths between one set of neurons and another using fetch_adjacencies().
The “source” and/or “target” neurons are selected using NeuronCriteria. Additional parameters allow you
to filter by connection strength or ROI. Two DataFrames are returned, for properties of pre/post synaptic
neurons and per-ROI connection strengths (for each pre-post pair)
'''
# Example: Fetch all downstream connections FROM a set of neurons
neuron_df, conn_df = fetch_adjacencies(
sources=[387023620, 387364605, 416642425], targets=None)
# Example: Fetch all upstream connections TO a set of neurons
neuron_df, conn_df = fetch_adjacencies(
sources=None, targets=[387023620, 387364605, 416642425])
# Example: Fetch all direct connections between a set of upstream neurons and downstream neurons
neuron_df, conn_df = fetch_adjacencies(sources=NC(type='Delta.*', regex=True),
targets=NC(type='PEN.*', regex=True))
# Print connections in descending order of strength
print(conn_df.sort_values('weight', ascending=False))
# ------------------------------------------
# Connection matrix
# ------------------------------------------
def get_all_connections():
connections_crit = NC()
neuron_df, conn_df = fetch_adjacencies(connections_crit, None)
print('got adjacencies!!!!!!!!!!!!!!!!!!!!!!')
conn_df.to_csv('all_connections.csv', index=False)
print('got csv file!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!')
