Added fixtures - #1419
Merged
Merged
Conversation
added 4 commits
August 24, 2026 13:37
Two gaps a new reader hits: - the home page said what the protocol is for and how it is built, but never which methods it implements. A reader had to click through the sidebar to discover TensorQTL, MASH, SuSiE, ColocBoost, cTWAS or S-LDSC. Adds a table of the seven analysis stages, with the workflow builder offered as the way to narrow them down. - Environment Setup took a reader from nothing to a working run on the committed fixtures, then stopped at the point they need most: swapping those paths for their own study. Adds a section covering the parameters most modules expect (--genoFile, --phenoFile, --covFile, --region-list, --cwd), the requirement that sample identifiers match across all three input files, and how to find a known-good parameter set from a module test. Parameter names verified against the notebooks rather than assumed: --phenoFile appears 171 times, --genoFile 141, --covFile 66, --region-list 60.
The FIXTURES entry for mnm_regression listed univariate_bvsr.rds and univariate_twas_weights.rds as inputs, with outputs empty. Both are what susie_twas produces -- they live under expected/ -- so the page told readers that fine-mapping consumes a fine-mapping result. Inputs are now the upstream products the test actually passes: protocol_example.genotype.chr22.bed from genotype_formatting protocol_example.pheno_manifest_context.tsv from phenotype_formatting example_covariates.tsv from covariate_preprocessing association_windows.bed from generalized_TADB and the two .rds files move to outputs. Note the covariate fixture is example_covariates.tsv, not protocol_example.covariates.tsv -- the convention would suggest the latter but it does not exist; taken from the covFile argument in test_mnm_regression.py. FXWF needed the same four keys, otherwise the Input section renders empty: it filters files by workflow, and these entries carry no data-wf because the workflow comes from the method chooser. Set to * so they show for every method. All six paths verified present on disk before writing, and the rendered bubble checked.
One module had no inputs listed and six had no outputs, so their I/O panels were half empty even though the fixtures were committed. Every path added is verified to exist on disk, and each was taken from what the module test actually passes or compares against. colocboost inputs from qtl_mini; expected/ colocboost.rds mash_posterior input from tests/fixtures/mash; three expected .rds ld_prune_reference genotype .bed and .ld_genotype.list; LD_pruned_variants rss_ld_sketch chr22 vcf and ld_blocks.bed; afreq and event_id tsv snRNAseq_preprocessing id_mapping.csv and seurat_ref_SE.rds RNA_calling fastq list and adapters; six expected outputs apa_calling 3UTR bed and gtf; four expected_* files FXWF gained the same keys, otherwise the sections render empty: it filters files by workflow, and these entries take their workflow from the method chooser rather than a data-wf attribute. Fixture layout is inconsistent across modules, which is why several of these were missed at first: some use an expected/ subdirectory, apa_calling uses an expected_* prefix in the module directory, and mash_posterior has no marker at all. Worth standardising. The two multivariate mnm_regression entries still show no expected output. No multicontext fixture is committed and none exists in output/ to subset; the step has no test either. Left empty rather than showing univariate results under a multivariate heading.
The two multivariate builder entries showed no expected output: no multicontext
fixture was committed and none existed in output/ to subset. Generated by running
the step on the committed toy data.
tests/fixtures/mnm_regression/expected/
protocol_example.ENSG00000283047.multicontext_bvsr.rds 334 KB
A genuine joint fit, not a reshaped univariate result: method is mvsusie and it
carries a jointContexts column the per-context object lacks.
Three errors in the documented command surfaced while getting it to run, all
recorded in the new expected/README.md with the working invocation:
- the parameter is --prior, not --mixture-prior
- --transpose-covariates True is required; without it the run fails with
No shared samples between phenotype and covariate file
- mnm cannot run alone; it needs qtl_dataset_construct+mnm
Caveats also in the README: the mixture prior was a choice between
mixture_prior.EE.prior.rds and prior.cov_ed.EE.rds, and no test drives the mnm
step, so this documents the expected shape rather than an asserted comparison.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Description
Updates fixture metadata and documentation across four commits and six files.
Workflow builder fixtures
mnm_regression, whose outputs were mistakenly listed as inputs.colocboost,RNA_calling,apa_calling,ld_prune_reference,mash_posterior,rss_ld_sketch, andsnRNAseq_preprocessing.FXWFso workflow-filtered panels render correctly.All 41 non-WIP modules now declare inputs and outputs.
METAL,qr_and_twas, andwatershedremain empty because they have no fixtures and are marked under development.Fixture layouts remain inconsistent and should be standardized separately.
Multicontext
mnm_regressionfixtureAdds a generated joint-fit fixture and regeneration notes:
The README records three corrections to the documented command:
--prior, not--mixture-prior.--transpose-covariates True.qtl_dataset_construct+mnm;mnmcannot run alone.The fixture uses
mixture_prior.EE.prior.rds. Because no test currently drivesmnm, it documents the expected shape rather than serving as an asserted comparison.Documentation
Reviewer note
Most changed lines come from long embedded notebook data maps; the substantive notebook changes are to
FIXTURESandFXWF. The branch name predates and does not describe this work.