Skip to content

Fixes - #216

Merged
dariarom94 merged 101 commits into
mainfrom
fixes
Aug 6, 2026
Merged

Fixes#216
dariarom94 merged 101 commits into
mainfrom
fixes

Conversation

@dariarom94

Copy link
Copy Markdown
Contributor

Describe your changes

Checklist before requesting a review

  • I have performed a self-review of my code

  • Check the correct box. Does this PR contain:

    • Breaking changes
    • New functionality
    • Major changes
    • Minor changes
    • Bug fixes
  • Proposed changes are described in the CHANGELOG.md

  • CI Tests succeed and look good!

dariarom94 and others added 30 commits July 19, 2026 23:22
The cellposev4 segmentation component exists but was not listed in any
run script, so it never ran. Add it to segmentation_methods alongside
cellpose (active in the seqeracloud scripts, commented in the local/test
scripts, matching the existing convention).

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Promote gene_efficiency_correction from an expression-correction method to
its own stage (methods_gene_efficiency_correction) that runs after
expression correction, offering two options: no_correction (pass-through)
and gene_efficiency_correction.

- New API src/api/comp_method_gene_efficiency_correction.yaml (input and
  output on file_spatial_corrected_counts).
- New components under methods_gene_efficiency_correction/; the moved
  gene_efficiency_correction now reads --input and preserves an upstream
  normalized_uncorrected layer.
- Remove gene_efficiency_correction from methods_expression_correction.
- run_benchmark: insert the gene_eff stage between expression correction and
  aggregate_spatial_data; it overwrites the output_correction state key so
  aggregate_spatial_data and the similarity metric need no rewiring and
  controls keep working. Add --gene_efficiency_correction_methods and alias
  the colliding no_correction dependency (gene_eff_no_correction).
- Update run scripts' method lists for the new stage.

Verified: viash config view, viash ns build (components + workflow, deps and
alias resolve), and viash test on both new components (output spec-conformant).

Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
dariarom94 and others added 29 commits August 2, 2026 11:09
The process_dataset crop fix (transform-agnostic crop_points_by_global_xy)
eliminated the out-of-bounds transcripts that motivated segger's in_bounds
exclusion + seg_orig_idx row_index remap; a full validation run on
re-processed Xenium + MERFISH confirmed n_oob == 0. Replace the exclusion
with a plain edge clamp matching basic/baysor/proseg, write all transcripts
so segger's row_index indexes tx_pd directly, and keep an n_oob == n_tx
coordinate-mismatch guard.

Script-only change (deploys via viash ns build + build/main regen, no image
rebuild). NOTES.md updated; the separate empty-bd validation-tile crash on
MERFISH is documented as still-open.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
…MERFISH)

Huge whole-section images (e.g. ABCA whole-brain MERFISH, ~83k x 102k px) place
the tissue well off-centre, so the image-centred crop window can miss it entirely
(mouse1_coronal kept 0 of 42M transcripts). Add a --tissue_centered_crop flag
(default false = historical image-centred behaviour) that centres the crop window
on the transcript density (median global x/y), clamped inside the image. Forwarded
through the process_datasets workflow fromState; enabled in the ABCA combine script.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
@dariarom94
dariarom94 merged commit 9049416 into main Aug 6, 2026
1 check passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant