def processing_pipeline(
experiment: starfish.Experiment,
fov_name: str,
n_processes: Optional[int] = None
) -> Tuple[starfish.ImageStack, starfish.IntensityTable]:
"""Process a single field of view of an experiment
Parameters
----------
experiment : starfish.Experiment
starfish experiment containing fields of view to analyze
fov_name : str
name of the field of view to process
n_processes : int
Returns
-------
starfish.IntensityTable :
decoded IntensityTable containing spots matched to the genes they are hybridized against
"""
print("Loading images...")
primary_image = experiment[fov_name].get_image(FieldOfView.PRIMARY_IMAGES)
all_intensities = list()
codebook = experiment.codebook
images = enumerate(experiment[fov_name].iterate_image_type(
FieldOfView.PRIMARY_IMAGES))
for image_number, primary_image in images:
print(f"Filtering image {image_number}...")
filter_kwargs = dict(in_place=True,
verbose=True,
n_processes=n_processes)
print("Applying Clip...")
clip1.run(primary_image, **filter_kwargs)
print("Applying Bandpass...")
bandpass.run(primary_image, **filter_kwargs)
print("Applying Gaussian Low Pass...")
glp.run(primary_image, **filter_kwargs)
print("Applying Clip...")
clip2.run(primary_image, **filter_kwargs)
print("Calling spots...")
spot_attributes = tlmpf.run(primary_image)
all_intensities.append(spot_attributes)
spot_attributes = IntensityTable.concatenate_intensity_tables(
all_intensities)
print("Decoding spots...")
decoded = codebook.decode_per_round_max(spot_attributes)
decoded = decoded[decoded["total_intensity"] > .025]
print("Processing complete.")
return primary_image, decoded
def test_take_max():
"""
Create two overlapping IntensityTables with differing number of spots and verify that
by concatenating them with the TAKE_MAX strategy we only include spots in the overlapping
section from the IntensityTable that had the most.
"""
it1 = create_intensity_table_with_coords(Area(min_x=0,
max_x=2,
min_y=0,
max_y=2),
n_spots=10)
it2 = create_intensity_table_with_coords(Area(min_x=1,
max_x=2,
min_y=1,
max_y=3),
n_spots=20)
concatenated = IntensityTable.concatenate_intensity_tables(
[it1, it2], overlap_strategy=OverlapStrategy.TAKE_MAX)
# The overlap section hits half of the spots from each intensity table, 5 from it1
# and 10 from i21. It2 wins and the resulting concatenated table should have all the
# spots from it2 (20) and 6 (one on the border) from it1 (6) for a total of 26 spots
assert concatenated.sizes[Features.AXIS] == 26
import os, fnmatch
from starfish import IntensityTable
filenames = []
listOfFiles = os.listdir('.')
pattern = "*.nc"
for entry in listOfFiles:
if fnmatch.fnmatch(entry, pattern):
filenames.append(entry)
# load files into memory and concat them
decoded_intensity_tables = map(IntensityTable.open_netcdf, filenames)
merged_table = IntensityTable.concatenate_intensity_tables(
decoded_intensity_tables)
merged_table.to_netcdf("merged_decoded_fovs.nc")