def default_celltypes(exclude = []):
priority_list = pymaid.get_annotated('mw brain simple priorities').name
priority_skids = [Celltype_Analyzer.get_skids_from_meta_meta_annotation(priority) for priority in priority_list]
# made the priority groups exclusive by removing neurons from lists that also in higher priority
override = priority_skids[0]
priority_skids_unique = [priority_skids[0]]
for i in range(1, len(priority_skids)):
skids_temp = list(np.setdiff1d(priority_skids[i], override))
priority_skids_unique.append(skids_temp)
override = override + skids_temp
# take all 'mw brain simple groups' skids (under 'mw brain simple priorities' meta-annotation)
# and remove skids that aren't in the appropriate priority_skids_unique level
priority_skid_groups = [list(pymaid.get_annotated(meta).name) for meta in priority_list]
skid_groups = []
for i in range(0, len(priority_skid_groups)):
group = []
for j in range(0, len(priority_skid_groups[i])):
skids_temp = pymaid.get_skids_by_annotation(pymaid.get_annotated(priority_skid_groups[i][j]).name)
skids_temp = list(np.intersect1d(skids_temp, priority_skids_unique[i])) # make sure skid in subgroup is set in correct priority list
# remove neurons in optional "exclude" list
if(len(exclude)>0):
skids_temp = list(np.setdiff1d(skids_temp, exclude))
group.append(skids_temp)
skid_groups.append(group)
# test skid counts for each group
#[len(x) for sublist in skid_groups for x in sublist]
# make list of lists of skids + their associated names
skid_groups = [x for sublist in skid_groups for x in sublist]
names = [list(pymaid.get_annotated(x).name) for x in priority_list]
names = [x for sublist in names for x in sublist]
names = [x.replace('mw brain ', '') for x in names]
# identify colors
colors = list(pymaid.get_annotated('mw brain simple colors').name)
# official order; note that it will have to change if any new groups are added
official_order = ['sensories', 'PNs', 'ascendings', 'PNs-somato', 'LNs', 'LHNs', 'FFNs', 'MBINs', 'KCs', 'MBONs', 'MB-FBNs', 'CNs', 'pre-dSEZs', 'pre-dVNCs', 'RGNs', 'dSEZs', 'dVNCs']
colors_names = [x.name.values[0] for x in list(map(pymaid.get_annotated, colors))] # use order of colors annotation for now
if(len(official_order)!=len(colors_names)):
print('warning: issue with annotations! Check "official_order" in Celltype_Analyzer.default_celltypes()')
# ordered properly and linked to colors
groups_sort = [np.where(x==np.array(official_order))[0][0] for x in names]
names = [element for _, element in sorted(zip(groups_sort, names))]
skid_groups = [element for _, element in sorted(zip(groups_sort, skid_groups))]
color_sort = [np.where(x.replace('mw brain ', '')==np.array(official_order))[0][0] for x in colors_names]
colors = [element for _, element in sorted(zip(color_sort, colors))]
data = pd.DataFrame(zip(names, skid_groups, colors), columns = ['name', 'skids', 'color'])
celltype_objs = list(map(lambda x: Celltype(*x), zip(names, skid_groups, colors)))
return(data, celltype_objs)
def get_skids_from_meta_meta_annotation(meta_meta, split=False, return_celltypes=False):
meta_annots = pymaid.get_annotated(meta_meta).name
annot_list = [list(pymaid.get_annotated(meta).name) for meta in meta_annots]
skids = [list(pymaid.get_skids_by_annotation(annots)) for annots in annot_list]
if(split==False):
skids = [x for sublist in skids for x in sublist]
return(skids)
if(split==True):
if(return_celltypes==True):
celltypes = [Celltype(meta_annots[i], skids[i]) for i in range(len(meta_annots))]
return(celltypes)
if(return_celltypes==False):
return(skids, meta_annots)
def get_skids_from_meta_annotation(meta, split=False, unique=True, return_celltypes=False):
annot_list = pymaid.get_annotated(meta).name
skids = [list(pymaid.get_skids_by_annotation(annots)) for annots in annot_list]
if(split==False):
skids = [x for sublist in skids for x in sublist]
if(unique==True):
skids = list(np.unique(skids))
return(skids)
if(split==True):
if(return_celltypes==True):
celltypes = [Celltype(annot_list[i], skids[i]) for i in range(len(annot_list))]
return(celltypes)
if(return_celltypes==False):
return(skids, annot_list)
def purge_unused_annotations(remote_instance=None, force=False, verbose=True):
if remote_instance in [None, 'source']:
remote_instance = source_project
if verbose:
print('Purging unused annotations from the source project.')
elif remote_instance == 'target':
remote_instance = target_project
if verbose:
print('Purging unused annotations from the target project.')
pid = remote_instance.project_id
tmp_neuron_skids = {2: 6401, 13: 120221, 38: 352154, 59: 665257}
if pid not in tmp_neuron_skids:
raise ValueError(
f'Need to specify a dummy neuron to use for project id {pid}')
tmp_skid = tmp_neuron_skids[pid]
all_annots = pymaid.get_annotation_list(remote_instance=remote_instance)
def add_escapes(s, chars_to_escape='()[]?+'):
for char in chars_to_escape:
s = s.replace(char, '\\' + char)
return s
def remove_escapes(s):
while '\\' in s:
s = s.replace('\\', '')
return s
for name in tqdm(all_annots.name):
name = add_escapes(name)
try:
count = len(
pymaid.get_annotated(name, remote_instance=remote_instance))
if count > 0:
pass #print('{:04d}'.format(count), name)
else:
print(name)
if not force and input('Purge me? [Y/n] ').lower() != 'y':
continue
pymaid.add_annotations(tmp_skid,
remove_escapes(name),
remote_instance=remote_instance)
pymaid.remove_annotations(tmp_skid,
remove_escapes(name),
remote_instance=remote_instance)
except:
print(name)
raise
def df_from_meta_annotation(key, filt=None):
print(f"Getting annotations under {key}:\n")
annot_df = pymaid.get_annotated(key)
series_ids = []
for annot_name in annot_df["name"]:
print("\t" + annot_name)
ids = pymaid.get_skids_by_annotation(annot_name.replace("*", "\*"))
if filt is not None:
name = filt(annot_name)
else:
name = annot_name
indicator = pd.Series(index=ids,
data=np.ones(len(ids), dtype=bool),
name=name,
dtype=bool)
series_ids.append(indicator)
print()
return pd.concat(series_ids, axis=1, ignore_index=False)
def test_get_annotated(self):
self.assertIsInstance(pymaid.get_annotated(
config_test.test_annotations[0], remote_instance=self.rm),
pd.DataFrame)
sc = get_single_class(classes[inds])
else:
if fill_unk:
sc = "unk"
else:
sc = ""
single_class.append(sc)
return single_class, all_class, n_class
# %% [markdown]
# ##
print("Loading annotations:\n")
start_instance()
annot_df = pymaid.get_annotated("mw neuron groups")
series_ids = []
for annot_name in annot_df["name"]:
print(annot_name)
ids = pymaid.get_skids_by_annotation(annot_name)
name = annot_name.replace("mw ", "")
name = name.replace(" ", "_")
indicator = pd.Series(index=ids,
data=np.ones(len(ids), dtype=bool),
name=name,
dtype=bool)
series_ids.append(indicator)
print()
# %% [markdown]
