Main Snakefile DAG should replicate the subworkflow DAG in an identical manner.

[rajwanir]

Description:

Some of the workflow logic and works often need to be redefined both within the subworkflow as well as the main snakefile. This sometimes results in redundancy. If a change to workflow rule selection logic is made, it needs to be modified in both places the main snakfile and the subworkflow for correct/desired DAG.

For example:

Checks on whether to execute contamination subworkflow and summarize it's results is referenced in both main snakefile and sample_qc subworkflow:

GwasQcPipeline/src/cgr_gwas_qc/workflow/Snakefile

Line 51 in e326cda

if sample_qc.use_contamination or cfg.config.user_files.bcf:

GwasQcPipeline/src/cgr_gwas_qc/workflow/sub_workflows/sample_qc.smk

Line 254 in e326cda

if use_contamination or (idat_intensity_retrieved and contamination_checked):

Similar issue was noted when adding a logic to skip plink ibd in PR #367 .

Solution:

Unless there comes a use case to implement a secondary rule selection in main snakefile, It might be preferable to have rule selection logics defined in the subworkflow and the main snakefile simply replicates the subworkflow logic in an identical manner. The main Snakefile would be a good place to select which subworklfows gets executed. But the DAG of the concerned rules for subworkflow and main snakefile should be identical for ease in maintainence.