fix pipeline

[dimalvovs]

This pull request updates the spacemarkers pipeline to improve reproducibility, flexibility, and gene selection logic. The most important changes include pinning all container images to a specific digest for reproducibility, updating function calls to use new snake_case naming conventions, and adding configurable parameters for ligand-receptor gene filtering and gene expression thresholds. Additionally, the logic for gene selection in the HD pipeline is now controlled by pipeline parameters.

Reproducibility and Container Management:

• All references to the spacemarkers container in Nextflow processes are now pinned to a specific image digest (sha256:9c06f8f...) instead of using floating tags. This ensures consistent environments across runs. [1] [2] [3] [4] [5]

Function Naming Consistency:

• Updated several R function calls in main.nf to use snake_case (e.g., get_spatial_features, get_spatial_parameters, find_all_hotspots, calculate_overlap_undirected, get_pairwise_interacting_genes) for consistency with updated codebase conventions. [1] [2]

Parameterization and Flexibility:

• Added new parameters use_ligand_receptor_genes (boolean) and good_gene_threshold (integer) to nextflow.config for controlling gene filtering behavior in the HD pipeline.

Gene Selection Logic in HD Pipeline:

• Refactored hdPipeline.R to use the new parameters: if use_ligand_receptor_genes is true, the analysis is restricted to ligand-receptor genes using the CellChat database; otherwise, the top good_gene_threshold most highly expressed genes are selected for analysis. [1] [2] [3]

Output Handling Improvements:

• The LRscores output in the SPACEMARKERS_HD process is now marked as optional, reflecting that it may not always be produced depending on pipeline parameters.

[dimalvovs]

upstream happy case tests pass

🚀 nf-test 0.9.2
https://www.nf-test.com
(c) 2021 - 2024 Lukas Forer and Sebastian Schoenherr


Test Workflow main.nf

  Test [2b1b8044] 'Just check samplesheet and read visium to adata' PASSED (39.283s)
  Test [5d40ee1a] 'Fail on empty samplesheet' PASSED (5.221s)
  Test [abcbcbcb] 'RCTD on Visium SD with local ref' PASSED (132.248s)
  Test [c2f6ac3f] 'RCTD on Multisample Visium SD with local ref' PASSED (453.398s)
  Test [2424d674] 'Full pipeline on Visium SD with remote ref' PASSED (290.981s)
  Test [b008168f] 'Full pipeline on Visium HD with remote ref' PASSED (199.744s)
  Test [c17e5da8] 'Visium SD with external deconvolution' PASSED (117.813s)

[Copilot]

Pull Request Overview

This PR refactors the HD pipeline to make ligand-receptor gene analysis configurable and updates container references to use SHA256 digests. It also standardizes function naming conventions to use snake_case.

• Adds configurable parameters for ligand-receptor gene filtering and gene expression thresholds
• Updates container references from tag-based to SHA256 digest-based pinning
• Refactors function names from camelCase to snake_case for consistency

Reviewed Changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 1 comment.

File	Description
nextflow/templates/hdPipeline.R	Refactored to use configurable parameters, moved ligand-receptor loading inside conditional block, and updated "good genes" logic to select top N genes
nextflow/nextflow.config	Added new parameters for ligand-receptor gene usage and gene threshold configuration
nextflow/main_hd.nf	Updated container to SHA256 digest and made LRscores output optional
nextflow/main.nf	Updated container to SHA256 digest and refactored function names to snake_case

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

The comment 'Use top "good" genes' is misleading. It should be updated to 'Select top N genes by expression' or similar to accurately reflect that this selects the top genes by total expression rather than filtering by a threshold.

Suggested change

	# Use top "good" genes
	# Select top N genes by total expression

[atuldeshpande]

This is the only one I did not test. Is this the right container address? A named version would look nicer?

[dimalvovs]

would look nicer, but would provide less reproducibility guarantees, voting to keep as is