Run FastQC and restore its citation text - #76
Conversation
The module was installed, configured and documented in output.md, but no subworkflow ever called it, so results/fastqc/ was never produced and the MultiQC report carried nothing but versions and the run summary. Wires FASTQC onto the pre-merge reads behind skip_preprocessing_qc, feeds its zip output to MultiQC, and uncomments the FastQC citation and bibliography entries that were disabled while the process was dormant.
|
nickp60
left a comment
There was a problem hiding this comment.
Hi @vinisalazar, thanks for getting this started! We had discussed a while ago about not manually calling fastqc, in favor of adding an option to run the nf-core subworkflow fastq_shortread_preprocess_qc. If you feel like we need it for v1, could you swap fastqc for that subworkflow?
A lazy efficient way to get this through for review could be to document how to run the subworkflow directly in the README with nextflow module run nf-core/fastq_shortreads_preprocess_qc syntax, if that can be used with subworkflows (I haven't tried it).
|
Hey Nick, thanks for the refresher. I'll have a go at adding the subworkflow, but if I feel I'm getting stuck, will look into that alternative of documenting that manually |
DATAPREP now runs FASTQ_SHORTREADS_PREPROCESS_QC per sequencing run, before run merging, and emits its logs and reports on a new multiqc_files channel. This replaces the hand-rolled FASTQC call in workflows/funcprofiler.nf. Adapter trimming, complexity filtering and host removal are exposed through taxprofiler-style params and are all off by default, so the only behaviour change out of the box is that FastQC now runs twice, on raw and on processed reads. The subworkflow's own final concatenation stays off because MERGE_RUNS already does that job.
The dev merge kept an allOf entry pointing at preprocessing_run_merging_options after dev had deleted the group itself, which failed schema_lint on nf-core#76 with "Subschema `preprocessing_run_merging_options` found in `allOf` but not in `$defs`". Drop the dangling ref, and the save_runmerged_reads param it described, which dev had already removed as unused.
The default test disabled FASTQC with ext.when = false, so docs/output.md advertised a fastqc/ directory the snapshot never contained. tests/.nftignore already ignored FastQC report contents, so dropping the gate records the paths without depending on unstable content. Runtime goes from ~170s to ~366s. Also note that the pipeline never checks db_path against the tool version, with the combinations exercised on the pilot cohort.
767848b to
b46d2aa
Compare
|
Hi @nickp60, sorry this is a big one, as the |
Module wasn't being called at all