Faster compute average intensities

[mgeplf]

Most of the speed up comes from not having to repeatedly create masks:

old: 1920s
new:
2min 26s
so about 13 times faster
10s, so about 192 times faster

Small change in computed densities:
(Pdb++) np.abs(result.to_numpy() - old.to_numpy()).max()
9.518734625513225e-08
Test code:

from atlas_densities.densities import fitting
from voxcell import VoxelData, RegionMap
from atlas_densities.densities import utils

annotation = VoxelData.load_nrrd('input-data/annotation_ccfv2_l23split_barrelsplit_validated.nrrd')

gene_voxeldata = {
    'gad67': VoxelData.load_nrrd('input-data/gene_gad67_correctednissl.nrrd'),
    'pv': VoxelData.load_nrrd('input-data/gene_pv_correctednissl.nrrd'),
    'sst': VoxelData.load_nrrd('input-data/gene_sst_correctednissl.nrrd'),
    'vip': VoxelData.load_nrrd('input-data/gene_vip_correctednissl.nrrd'),
}
gids = { 'gad67': '479', 'pv': '868', 'sst': '1001', 'vip': '77371835', }
slices = utils.load_json('input-data/realigned_slices.json')
gene_marker_volumes = {
    gene: {
        "intensity": gene_data.raw,
        "slices": slices[gids[gene]],  # list of integer slice indices
    }
    for (gene, gene_data) in gene_voxeldata.items()
}
region_map = RegionMap.load_json('input-data/hierarchy_ccfv2_l23split_barrelsplit.json')
hierarchy_info = utils.get_hierarchy_info(region_map, root='root')

res = fitting.compute_average_intensities(annotation.raw, gene_marker_volumes, hierarchy_info, region_map)

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 97.01493% with 2 lines in your changes are missing coverage. Please review.

❗ No coverage uploaded for pull request base (main@5b08942). Click here to learn what that means.

Files	Patch %	Lines
atlas_densities/densities/fitting.py	97.01%	2 Missing ⚠️

Additional details and impacted files

@@           Coverage Diff           @@
##             main      #73   +/-   ##
=======================================
  Coverage        ?   97.67%           
=======================================
  Files           ?       22           
  Lines           ?     1508           
  Branches        ?        0           
=======================================
  Hits            ?     1473           
  Misses          ?       35           
  Partials        ?        0           

Flag	Coverage Δ
pytest	97.67% <97.01%> (?)

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.