CodyCBakerPhD
commented
11 months ago Reference script
```python
from pathlib import Path
from collections import defaultdict
from typing import Union, List
import h5py
from natsort.natsort import natsorted
SKIP_KEYS = ["timeseries", "corrected", "original"]
ABBREVIATE_KEYS = ["imaging_plane_1"]
def _recurse_structure(dataset_list: list, group: h5py.Group, key: Union[str, None]):
if isinstance(group[key], h5py.Group):
next_group = group[key]
for next_key in next_group:
if next_key in SKIP_KEYS:
continue
if next_key in ABBREVIATE_KEYS:
dataset_list.append(next_group.name)
continue
_recurse_structure(dataset_list=dataset_list, group=next_group, key=next_key)
else:
dataset_list.append(group[key].name) # Is the full path within the HDF5 file, including '/'
def _find_all_datasets(file: h5py.File) -> List[str]:
dataset_list = list()
for key in file:
_recurse_structure(dataset_list=dataset_list, group=file, key=key)
return dataset_list
base_path = Path("G:/visual-coding/ophys_experiment_data")
v1_nwbfiles = list(base_path.rglob("*.nwb"))
datasets_per_file = defaultdict(list)
for v1_nwbfile in v1_nwbfiles:
file = h5py.File(name=v1_nwbfile, mode="r")
datasets_per_file[v1_nwbfile] = _find_all_datasets(file=file)
all_datasets = list()
for datasets in datasets_per_file.values():
all_datasets.extend(datasets)
unique_datasets = natsorted(list(set(all_datasets)))
# import json
# print(json.dumps(unique_datasets, indent=4))
unique_counts = {key: 0 for key in unique_datasets}
for datasets in datasets_per_file.values():
for dataset in datasets:
unique_counts[dataset] += 1
```
Evaluate how consistent various, especially behavioral, neurodata objects are across the datasets