Merge pull request #265 from EBI-Metagenomics/rawr-functional

Add option for functional analysis to RawR

Author: timrozday-mgnify

.gitignore

@@ -29,3 +29,5 @@ slurm-dev-environment/fs/nfs/ftp/public/databases/metagenomics/mgnify_results/*
 genomes/temp/*
 slurm-dev-environment/fs/nfs/public/tests/assembly_v6_output/ERP106708/MGYS*
 slurm-dev-environment/fs/nfs/public/tests/amplicon_v6_output/dwca/*
+.claude/*
+CLAUDE.md

.talismanrc

@@ -172,7 +172,7 @@ fileignoreconfig:
   checksum: ac644b851d31913ca9328d5a83516146c4c76874201df0a2d7086df6b60435c8
 
 - filename: workflows/tests/test_analysis_rawreads_study_flow.py
-  checksum: 0f7255bdd7303269337a7d9be17d5246bfc32980697ba4135f54be201d13180a
+  checksum: 211d4cfcd3e1b0dbe6e8b92683f6160661c50a17429d470ea0651daa41f5c3f8
 
 - filename: workflows/flows/upload_assembly.py
   allowed_patterns: [key]

workflows/flows/analyse_study_tasks/raw_reads/run_rawreads_pipeline_via_samplesheet.py

@@ -44,6 +44,7 @@ def run_rawreads_pipeline_via_samplesheet(
     rawreads_analysis_ids: List[Union[str, int]],
     workdir: Optional[Path],
     outdir: Optional[Path],
+    functional_analysis: bool = False,
 ):
     if workdir is None:
         workdir = (
@@ -94,6 +95,7 @@ def run_rawreads_pipeline_via_samplesheet(
             ("--outdir", nextflow_outdir),
             EMG_CONFIG.slurm.use_nextflow_tower and "-with-tower",
             EMG_CONFIG.rawreads_pipeline.has_fire_access and "--use_fire_download",
+            ("--skip_functional", "false" if functional_analysis else "true"),
             ("-work-dir", nextflow_workdir),
             ("-ansi-log", "false"),
         ]

workflows/flows/analysis_rawreads_study.py

@@ -103,6 +103,10 @@ class AnalyseStudyInput(RunInput):
             ENALibraryStrategyPolicy.ONLY_IF_CORRECT_IN_ENA,
             description="Optionally treat read-runs with incorrect library strategy metadata as raw-reads.",
         )
+        functional_analysis: bool = Field(
+            False,
+            description="Enable functional analysis in the raw-reads pipeline.",
+        )
         webin_owner: Optional[str] = Field(
             None,
             description="Webin ID of study owner, if data is private. Can be left as None, if public.",
@@ -208,7 +212,11 @@ class AnalyseStudyInput(RunInput):
             f"Working on raw-reads analyses: {analyses_chunk[0]}-{analyses_chunk[-1]}"
         )
         run_rawreads_pipeline_via_samplesheet(
-            mgnify_study, analyses_chunk, study_workdir, study_outdir
+            mgnify_study,
+            analyses_chunk,
+            study_workdir,
+            study_outdir,
+            functional_analysis=analyse_study_input.functional_analysis,
         )
 
     merge_study_summaries(