def atlas_averages_annotationOnly(new_metadata, new_counttable, **kwargs):
n_pcs = kwargs['n_pcs']
atlas = kwargs['atlas']
cell_type_names = kwargs['CT_lut']
#instantiate class
sa = northstar.Averages(
atlas=atlas,
n_cells_per_type=kwargs['weights_atlas_cells'],
n_features_per_cell_type=kwargs['n_features_per_cell_type'],
n_features_overdispersed=kwargs['n_features_overdispersed'],
n_pcs=n_pcs,
n_neighbors=kwargs['n_neighbors'],
n_neighbors_out_of_atlas=kwargs['n_neighbors_out_of_atlas'],
distance_metric='correlation',
threshold_neighborhood=kwargs['threshold_neighborhood'],
clustering_metric='cpm',
resolution_parameter=kwargs['resolution_parameter'],
normalize_counts=True,
)
sa.fit(new_counttable)
n_fixed = len(sa.cell_types)
idx = list(sa.cell_types) + list(new_counttable.columns)
annotdf = pd.DataFrame(index=idx, columns=['new_membership', 'class'])
idx = new_counttable.columns
new_metadata.loc[idx, 'new_class'] = sa.membership
new_metadata['new_class_renamed'] = [
cell_type_names[f] if f in cell_type_names.keys() else 'NewClass_' +
"{0:0=2d}".format(int(f) - n_fixed + 1) if (f.isdigit() == True) else f
for f in new_metadata['new_class']
]
annotdf.loc[idx, 'new_membership'] = new_metadata.loc[idx,
'new_class_renamed']
annotdf.loc[annotdf[:n_fixed].index,
'new_membership'] = annotdf.index[:n_fixed].map(
cell_type_names)
atlastypes = list(
np.sort(annotdf.loc[annotdf[:n_fixed].index, 'new_membership']))
newtypes = list(
set(new_metadata['new_class_renamed']).difference(atlastypes))
celltypes = atlastypes + list(np.sort(newtypes))
class_lut = dict(zip(celltypes, list(range(1, len(celltypes) + 1))))
annotdf['class'] = annotdf['new_membership'].map(class_lut)
return annotdf
'class': 'subsample',
'n_atlas': ns.n_atlas,
}
ress.append(resd)
print('Run northstar (averages)')
for aname, atlas in atlases.items():
if not aname.endswith('avg'):
continue
print('Atlas: {:}'.format(aname))
t0 = time.time()
ns = northstar.Averages(
atlas,
n_features_per_cell_type=50,
n_features_overdispersed=500,
resolution_parameter=0.005,
n_neighbors=30,
#n_neighbors_external=5,
#external_neighbors_mutual=True,
)
ns.fit(adata_tgt)
t1 = time.time()
t = t1 - t0
ct_orig = ns.new_data.obs['CellType'].astype(str)
gof = (ct_orig == ns.membership).mean()
identity = ct_orig.to_frame()
identity['northstar_assignment'] = ns.membership
vs = ns.embed('umap')
def atlas_averages_to_tsnedf(new_metadata, new_counttable, **kwargs):
savedir = kwargs['savedir']
date = kwargs['timestamp']
n_pcs = kwargs['n_pcs']
atlas = kwargs['atlas']
cell_type_names = kwargs['CT_lut']
#instantiate class
sa = northstar.Averages(
atlas=atlas,
n_cells_per_type=kwargs['weights_atlas_cells'],
n_features_per_cell_type=kwargs['n_features_per_cell_type'],
n_features_overdispersed=kwargs['n_features_overdispersed'],
n_pcs=n_pcs,
n_neighbors=kwargs['n_neighbors'],
n_neighbors_out_of_atlas=kwargs['n_neighbors_out_of_atlas'],
distance_metric='correlation',
threshold_neighborhood=kwargs['threshold_neighborhood'],
clustering_metric='cpm',
resolution_parameter=kwargs['resolution_parameter'],
normalize_counts=True,
)
sa.fit(new_counttable)
# add new membership to metadata
idx = new_counttable.columns
n_fixed = len(sa.cell_types)
new_metadata.loc[idx, 'new_class'] = sa.membership
new_metadata['new_class_renamed'] = [
cell_type_names[f] if f in cell_type_names.keys() else 'NewClass_' +
"{0:0=2d}".format(int(f) - n_fixed + 1) if (f.isdigit() == True) else f
for f in new_metadata['new_class']
]
# unweighted PCA
cols = list(sa.cell_types) + list(new_counttable.columns)
feature_selected_matrix = pd.DataFrame(index=sa.features_selected,
columns=cols,
data=sa.matrix)
normal_PCA, distance_matrix = unweighted_PCA(feature_selected_matrix,
n_pcs)
# perform tSNE
tsnedf = perform_tSNE(normal_PCA, 20)
tsnedf.rename(index=str, columns={0: 'Dim1', 1: 'Dim2'}, inplace=True)
tsnedf.loc[idx, 'new_membership'] = new_metadata.loc[idx,
'new_class_renamed']
tsnedf.loc[tsnedf[:n_fixed].index,
'new_membership'] = tsnedf.index[:n_fixed].map(cell_type_names)
# write params to json in new folder with date timestamp
output_file = savedir + date + '/annotation_parameters_' + atlas + '_CellAtlasAverages_' + date + '.json'
if not os.path.exists(os.path.dirname(output_file)):
try:
os.makedirs(os.path.dirname(output_file))
except OSError as exc:
if exc.errno != errno.EEXIST:
raise
with open(output_file, 'w') as file:
file.write(json.dumps(kwargs))
file.close()
# save feature matrix for later reference, e.g. making dotplots
feature_selected_matrix.to_csv(savedir + date +
'/feature_selected_matrix_' + date + '.csv')
atlastypes = list(
np.sort(tsnedf.loc[tsnedf[:n_fixed].index, 'new_membership']))
newtypes = list(
set(new_metadata['new_class_renamed']).difference(atlastypes))
celltypes = atlastypes + list(np.sort(newtypes))
class_lut = dict(zip(celltypes, list(range(1, len(celltypes) + 1))))
tsnedf['class'] = tsnedf['new_membership'].map(class_lut)
return tsnedf, celltypes, distance_matrix