New cyto tool: create cell locations file #257
Conversation
python \
pycytominer/cyto_utils/cell_locations_cmd.py \
--input_parquet_file pycytominer/tests/test_data/cell_locations_example_data/load_data_with_illum_subset.parquet \
--sqlite_file pycytominer/tests/test_data/cell_locations_example_data/BR00126114_subset.sqlite \
--output_parquet_file \
~/Desktop/load_data_with_illum_and_cell_location_subset.parquet |
Codecov Report
@@ Coverage Diff @@
## master #257 +/- ##
==========================================
- Coverage 95.71% 95.47% -0.25%
==========================================
Files 53 57 +4
Lines 2826 3048 +222
==========================================
+ Hits 2705 2910 +205
- Misses 121 138 +17
Flags with carried forward coverage won't be shown. Click here to find out more.
... and 3 files with indirect coverage changes 📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more |
|
This can be run on command line using metadata_input="s3://cellpainting-gallery/test-cpg0016-jump/source_4/workspace/load_data_csv/2021_08_23_Batch12/BR00126114/load_data_with_illum_subset.parquet"
single_single_cell_input="s3://cellpainting-gallery/test-cpg0016-jump/source_4/workspace/backend/2021_08_23_Batch12/BR00126114/BR00126114_subset.sqlite"
augmented_metadata_output="s3://cellpainting-gallery/test-cpg0016-jump/source_4/workspace/load_data_csv/2021_08_23_Batch12/BR00126114/load_data_with_illum_and_cell_location_subset.parquet"
python \
pycytominer/cyto_utils/cell_locations_cmd.py \
--metadata_input ${metadata_input} \
--single_cell_input ${single_single_cell_input} \
--augmented_metadata_output ${augmented_metadata_output} \
add_cell_location |
|
Currently, one of the test fixtures ( For now, import boto3
from typing import List
from moto import mock_s3
from moto.server import ThreadedMotoServer
# moto patterns are from
# https://github.com/cytomining/CytoTable/blob/415f2ecd94b66c979bfda595448264089f9a1c80/tests/conftest.py
# credits: @gwaygenomics and @d33b
@pytest.fixture(scope="session", name="s3_session")
def fixture_s3_session() -> boto3.session.Session:
"""
Yield a mocked boto session for s3 tests.
"""
# start a moto server for use in testing
server = ThreadedMotoServer()
server.start()
with mock_s3():
yield boto3.session.Session()
@pytest.fixture(name="example_local_sources")
def fixture_example_local_sources() -> List[pathlib.Path]:
"""
Provide a list of example sources
"""
return [
pathlib.Path(__file__).parent.parent
/ "test_data"
/ "cell_locations_example_data"
/ "BR00126114_subset.sqlite",
pathlib.Path(__file__).parent.parent
/ "test_data"
/ "cell_locations_example_data"
/ "load_data_with_illum_subset.parquet",
]
@pytest.fixture()
def example_s3_endpoint(
s3_session: boto3.session.Session,
example_local_sources: List[pathlib.Path],
) -> str:
"""
Create a mocked bucket which includes example sources
"""
# s3 is a fixture defined above that yields a boto3 s3 client.
# Feel free to instantiate another boto3 S3 client -- Keep note of the region though.
endpoint_url = "http://localhost:5000"
bucket_name = "example"
# create s3 client
s3_client = s3_session.client("s3", endpoint_url=endpoint_url)
# create a bucket for content to land in
s3_client.create_bucket(Bucket=bucket_name)
# upload each example file to the mock bucket
for source_path in example_local_sources:
s3_client.upload_file(
Filename=str(source_path),
Bucket=bucket_name,
Key=source_path.name,
)
# return endpoint url for use in testing
return endpoint_url
@pytest.fixture(name="single_cell_input_file_s3")
def fixture_single_cell_input_file_s3(example_s3_endpoint) -> str:
"""
Provide a single cell input file for cell_locations test data
"""
return f"{example_s3_endpoint}/example/BR00126114_subset.sqlite" |
|
Meanwhile, @Arkkienkeli, can you inspect the parquet file attached in #257 (comment) to see if this kind of output works for you (for DeepProfiler) instead of creating a locations file? That is, if such a (parquet) file were available, would you no longer need to create a locations file? See the logic in https://github.com/shntnu/pycytominer/blob/bc2f01dfa0252bba5f3aa04cd8488f8dc7fe5de9/pycytominer/cyto_utils/cell_locations.py#L14-L38 to figure out what's happening |
|
Hi @shntnu, In the example you shared, So either we update DeepProfiler or I will need to convert those files into our format. |
Thank you so much for the detailed review -- I learned a lot! I've resolved all your comments except this one, related to
I skimmed the SingleCells class, and my initial impression was that the overlaps were fairly minimal. But it's possible that refactoring For example, some of these Move to a utils module:
Move to
|
There was a problem hiding this comment.
Nice work! Thank you for making changes and acknowledging my comments @shntnu ! I've provided some additional comments and would like to request further changes. I especially feel the example in the readme and anonymous S3 access items should be addressed (or documented).
Responding to your comment:
"But it's possible that refactoring SingleCells would expose more reusable functionality that CellLocations could benefit from."
Thank you for addressing this! Based on what you mentioned I feel this may be reaching the limits or out of scope for this PR. Perhaps this is a new issue for the project to explore later on?
Co-authored-by: Dave Bunten <[email protected]>
Your feedback was really motivating!
All set now |
|
@d33bs If there's nothing further, please go ahead and merge and we'll launch a bunch of jobs that are waiting on this. Thank you once again for all the feedback along the way! |
modified from EmbeddedArtistry
Description
This introduces a new tool to append X, Y locations of all cells in an image to each row of a LoadData CSV file. This will make it a lot easier for deep learning feature extraction workflows to do their thing (e.g. DeepProfiler will no longer need to download single-cell SQLite files to extract locations, etc.)
What is the nature of your change?
Checklist
Please ensure that all boxes are checked before indicating that a pull request is ready for review.