Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Improve omero get #3

Merged
merged 11 commits into from
Oct 14, 2024
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
118 changes: 0 additions & 118 deletions tools/omero/omero_get.xml

This file was deleted.

113 changes: 113 additions & 0 deletions tools/omero/omero_get_id.py
Original file line number Diff line number Diff line change
@@ -0,0 +1,113 @@
import argparse
import csv
import json

import ezomero as ez


def get_ids_ezo(user, pws, host, port, final_obj_type, parent_obj_type, parent_id=None, tsv_file="id_list.tsv"):
# Function to write tabular file from the ezomero output
def write_ids_to_tsv(data, header):
with open(tsv_file, 'a+', newline='') as f:
f.seek(0)
is_empty = f.tell() == 0 # Check if file is empty
writer = csv.writer(f, delimiter='\t')
if is_empty:
writer.writerow([header]) # Write the header
for item in data:
writer.writerow([item]) # Write each ID
try:
with ez.connect(user, pws, "", host, port, secure=True) as conn:
if final_obj_type == "Project":
proj_ids = ez.get_project_ids(conn)
write_ids_to_tsv(proj_ids, "Project IDs")
return proj_ids
elif final_obj_type == "Dataset":
args = {'project': None}
if parent_obj_type == "Project":
args['project'] = parent_id
ds_ids = ez.get_dataset_ids(conn, **args)
write_ids_to_tsv(ds_ids, "Dataset IDs")
return ds_ids
elif final_obj_type == "Image":
args = {
'project': None,
'dataset': None,
'plate': None,
'well': None
}
if parent_obj_type == "Project":
args['project'] = parent_id
elif parent_obj_type == "Dataset":
args['dataset'] = parent_id
elif parent_obj_type == "Plate":
args['plate'] = parent_id
elif parent_obj_type == "Well":
args['well'] = parent_id
elif parent_obj_type != "All":
raise ValueError("Object set as parent_obj_type is not compatible")
ds_ims = ez.get_image_ids(conn, **args)
write_ids_to_tsv(ds_ims, "Image IDs")
return ds_ims
elif final_obj_type == "Annotation":
map_annot_ids = ez.get_map_annotation_ids(conn, parent_obj_type, parent_id)
write_ids_to_tsv(map_annot_ids, "Annotation IDs")
return map_annot_ids
elif final_obj_type == "Tag":
tag_ids = ez.get_tag_ids(conn, parent_obj_type, parent_id)
write_ids_to_tsv(tag_ids, "Tag IDs")
return tag_ids
elif final_obj_type == "Roi":
roi_ids = ez.get_roi_ids(conn, parent_id)
write_ids_to_tsv(roi_ids, "ROI IDs")
return roi_ids
elif final_obj_type == "Table":
file_ann_ids = ez.get_file_annotation_ids(conn, parent_obj_type, parent_id)
write_ids_to_tsv(file_ann_ids, "Table IDs")
return file_ann_ids
else:
raise ValueError(f"Unsupported object type: {final_obj_type}")

except Exception as e:
print(f"Connection error: {str(e)}")
return None


# Argument parsing
if __name__ == "__main__":
parser = argparse.ArgumentParser(description="Fetch OMERO object IDs as TSV from parent object.")
parser.add_argument("--credential-file", dest="credential_file", type=str,
required=True, help="Credential file (JSON file with username and password for OMERO)")
parser.add_argument('--host', required=True,
help="Host server address.")
parser.add_argument('--port', required=True, type=int,
help='OMERO port')
parser.add_argument('--final_obj_type', required=True,
help="Type of object to fetch ID: Project, Dataset, Image, Annotation, Tag, Roi, or Table.")
parser.add_argument('--parent_obj_type', required=True,
help="Type of object from which you fetch IDs: Project, Dataset, Plate, Well, Image (or 'All' if you want to get all objects).")
parser.add_argument('--parent_id', required=False, type=int,
help="ID of the OMERO object in `--parent_obj_type`, not required if you used `--parent_obj_type All`.")
parser.add_argument('--tsv_file', default='id_list.tsv', required=True,
help="Output TSV file path.")
args = parser.parse_args()

if args.parent_id is None and args.parent_obj_type != "All":
raise ValueError("ID is only optional is you use `--parent_obj_type All`")

if args.final_obj_type == "Roi" and args.parent_obj_type != "Image":
raise ValueError("Roi IDs can only be retrived from images, use `--parent_obj_type Image`")

if args.parent_obj_type == "All" and args.final_obj_type not in ["Image", "Dataset", "Project"]:
raise ValueError("Only Images, Datasets and Projects is compatible with `--parent_obj_type All`")

with open(args.credential_file, 'r') as f:
crds = json.load(f)

# Call the main function to get the object and save it as a TSV
get_ids_ezo(user=crds['username'], pws=crds['password'], host=args.host,
port=args.port,
final_obj_type=args.final_obj_type,
parent_obj_type=args.parent_obj_type,
parent_id=args.parent_id,
tsv_file=args.tsv_file)
Loading