Merge pull request #132 from Clinical-Genomics/add_nallo

Add nallo
Clinical-Genomics · Feb 12, 2025 · 2fc375a · 2fc375a
2 parents 07e0cec + 0bd12a2
commit 2fc375a
Show file tree

Hide file tree

Showing 10 changed files with 669 additions and 6 deletions.
diff --git a/cg_hermes/cli/export.py b/cg_hermes/cli/export.py
@@ -11,6 +11,7 @@
 from cg_hermes.config.mip_dna import MIP_DNA_TAGS
 from cg_hermes.config.mip_rna import MIP_RNA_TAGS
 from cg_hermes.config.mutant import MUTANT_COMMON_TAGS
+from cg_hermes.config.nallo import NALLO_TAGS
 from cg_hermes.config.raredisease import RAREDISEASE_TAGS
 from cg_hermes.config.rnafusion import RNAFUSION_TAGS
 from cg_hermes.config.taxprofiler import TAXPROFILER_TAGS
@@ -68,6 +69,10 @@ class OutputFormat(StrEnum):
         "header": ["Mutant tags", "Mandatory", "HK tags", "Used by"],
         "tags": MUTANT_COMMON_TAGS,
     },
+    Workflow.NALLO: {
+        "header": ["Nallo tags", "Mandatory", "HK tags", "Used by"],
+        "tags": NALLO_TAGS,
+    },
     Workflow.RAREDISEASE: {
         "header": ["Raredisease tags", "Mandatory", "HK tags", "Used by"],
         "tags": RAREDISEASE_TAGS,

diff --git a/cg_hermes/cli/validate.py b/cg_hermes/cli/validate.py
@@ -11,6 +11,7 @@
 from cg_hermes.config.mip_dna import MIP_DNA_TAGS
 from cg_hermes.config.mip_rna import MIP_RNA_TAGS
 from cg_hermes.config.mutant import MUTANT_COMMON_TAGS
+from cg_hermes.config.nallo import NALLO_TAGS
 from cg_hermes.config.raredisease import RAREDISEASE_TAGS
 from cg_hermes.config.rnafusion import RNAFUSION_TAGS
 from cg_hermes.config.taxprofiler import TAXPROFILER_TAGS
@@ -73,6 +74,8 @@ def validate_tags_cmd(workflow: Workflow) -> None:
         tag_map = FLUFFY_COMMON_TAGS
     elif workflow == Workflow.MUTANT:
         tag_map = MUTANT_COMMON_TAGS
+    elif workflow == Workflow.NALLO:
+        tag_map = NALLO_TAGS
     elif workflow == Workflow.RAREDISEASE:
         tag_map = RAREDISEASE_TAGS
     elif workflow == Workflow.RNAFUSION:

diff --git a/cg_hermes/config/nallo.py b/cg_hermes/config/nallo.py
@@ -0,0 +1,290 @@
+"""Tags that are defined in deliverables file mapped to tags used in CG
+
+The tag sets that exists in all files are set to mandatory. Tag sets that exists in < 4 deliverables are not
+mandatory by default. However, the tags that are available to a particular analysis is mandatory for that analysis.
+"""
+
+from cg_hermes.config.nextflow import NEXTFLOW_TAGS
+from cg_hermes.constants.tags import (
+    AlignmentTags,
+    AnalysisTags,
+    BioinfoToolsTags,
+    FamilyTags,
+    NextflowTags,
+    NalloTags,
+    ReportTags,
+    VariantTags,
+    UsageTags,
+)
+
+NALLO_COMMON_TAGS = {
+    frozenset(["alignment", "alignment_haplotags"]): {
+        "tags": [AlignmentTags.BAM, NalloTags.HAPLOTAGS],
+        "is_mandatory": True,
+        "used_by": [UsageTags.SCOUT, UsageTags.CLINICAL_DELIVERY, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["alignment", "alignment_haplotags_index"]): {
+        "tags": [AlignmentTags.BAM_INDEX, NalloTags.HAPLOTAGS],
+        "is_mandatory": True,
+        "used_by": [UsageTags.SCOUT, UsageTags.CLINICAL_DELIVERY, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["assembly", "summary_hap1"]): {
+        "tags": [NalloTags.HAP1, NalloTags.ASSEMBLY, NalloTags.ASSEMBLY_SUMMARY],
+        "is_mandatory": True,
+        "used_by": [UsageTags.CLINICAL_DELIVERY, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["assembly", "summary_hap2"]): {
+        "tags": [NalloTags.HAP2, NalloTags.ASSEMBLY, NalloTags.ASSEMBLY_SUMMARY],
+        "is_mandatory": True,
+        "used_by": [UsageTags.CLINICAL_DELIVERY, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["assembly", "assembly_hap1_mapped"]): {
+        "tags": [AlignmentTags.BAM, NalloTags.HAP1, NalloTags.ASSEMBLY],
+        "is_mandatory": True,
+        "used_by": [UsageTags.CLINICAL_DELIVERY, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["assembly", "assembly_hap1_mapped_index"]): {
+        "tags": [AlignmentTags.BAM_INDEX, NalloTags.HAP1, NalloTags.ASSEMBLY],
+        "is_mandatory": True,
+        "used_by": [UsageTags.CLINICAL_DELIVERY, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["assembly", "assembly_hap2_mapped"]): {
+        "tags": [AlignmentTags.BAM, NalloTags.HAP2, NalloTags.ASSEMBLY],
+        "is_mandatory": True,
+        "used_by": [UsageTags.CLINICAL_DELIVERY, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["assembly", "assembly_hap2_mapped_index"]): {
+        "tags": [AlignmentTags.BAM_INDEX, NalloTags.HAP2, NalloTags.ASSEMBLY],
+        "is_mandatory": True,
+        "used_by": [UsageTags.CLINICAL_DELIVERY, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["assembly", "assembly_hap1_mapped"]): {
+        "tags": [AnalysisTags.BED, NalloTags.HAP1, BioinfoToolsTags.MODKIT_PILEUP],
+        "is_mandatory": True,
+        "used_by": [UsageTags.CLINICAL_DELIVERY, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["assembly", "assembly_hap1_mapped_index"]): {
+        "tags": [AnalysisTags.BED_INDEX, NalloTags.HAP1, BioinfoToolsTags.MODKIT_PILEUP],
+        "is_mandatory": True,
+        "used_by": [UsageTags.CLINICAL_DELIVERY, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["summary_counts", "hap1"]): {
+        "tags": [AnalysisTags.BED, NalloTags.HAP1, BioinfoToolsTags.MODKIT_PILEUP],
+        "is_mandatory": True,
+        "used_by": [UsageTags.CLINICAL_DELIVERY, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["summary_counts", "hap1_index"]): {
+        "tags": [AnalysisTags.BED_INDEX, NalloTags.HAP1, BioinfoToolsTags.MODKIT_PILEUP],
+        "is_mandatory": True,
+        "used_by": [UsageTags.CLINICAL_DELIVERY, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["summary_counts", "hap2"]): {
+        "tags": [AnalysisTags.BED, NalloTags.HAP2, BioinfoToolsTags.MODKIT_PILEUP],
+        "is_mandatory": True,
+        "used_by": [UsageTags.CLINICAL_DELIVERY, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["summary_counts", "hap2_index"]): {
+        "tags": [AnalysisTags.BED_INDEX, NalloTags.HAP2, BioinfoToolsTags.MODKIT_PILEUP],
+        "is_mandatory": True,
+        "used_by": [UsageTags.CLINICAL_DELIVERY, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["summary_counts", "ungrouped"]): {
+        "tags": [AnalysisTags.BED, NalloTags.UNGROUPED, BioinfoToolsTags.MODKIT_PILEUP],
+        "is_mandatory": True,
+        "used_by": [UsageTags.CLINICAL_DELIVERY, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["summary_counts", "ungrouped_index"]): {
+        "tags": [AnalysisTags.BED_INDEX, NalloTags.UNGROUPED, BioinfoToolsTags.MODKIT_PILEUP],
+        "is_mandatory": True,
+        "used_by": [UsageTags.CLINICAL_DELIVERY, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["qc_bam", "mosdepth_d4"]): {
+        "tags": [AnalysisTags.COVERAGE, ReportTags.D4],
+        "is_mandatory": True,
+        "used_by": [UsageTags.SCOUT, UsageTags.CLINICAL_DELIVERY, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset({"multiqc", "multiqc-html"}): {
+        "is_mandatory": True,
+        "tags": [ReportTags.MULTIQC_HTML],
+        "used_by": [UsageTags.SCOUT, UsageTags.CLINICAL_DELIVERY, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset({"multiqc", "multiqc-json"}): {
+        "is_mandatory": True,
+        "tags": [ReportTags.MULTIQC_JSON],
+        "used_by": [UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["pedigree", "pedigree_fam"]): {
+        "tags": [FamilyTags.PEDIGREE],
+        "is_mandatory": True,
+        "used_by": [UsageTags.CLINICAL_DELIVERY, UsageTags.SCOUT, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["somalier", "relate_html"]): {
+        "tags": [BioinfoToolsTags.SOMALIER, NalloTags.RELATE_HTML],
+        "is_mandatory": True,
+        "used_by": [UsageTags.CLINICAL_DELIVERY, UsageTags.SCOUT, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["somalier", "relate_pairs"]): {
+        "tags": [BioinfoToolsTags.SOMALIER, NalloTags.RELATE_PAIRS],
+        "is_mandatory": True,
+        "used_by": [UsageTags.SCOUT, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["somalier", "relate_pairs"]): {
+        "tags": [BioinfoToolsTags.SOMALIER, NalloTags.RELATE_PAIRS],
+        "is_mandatory": True,
+        "used_by": [UsageTags.SCOUT, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["somalier", "relate_samples"]): {
+        "tags": [BioinfoToolsTags.SOMALIER, NalloTags.RELATE_SAMPLES],
+        "is_mandatory": True,
+        "used_by": [UsageTags.SCOUT, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["deepvariant", "report"]): {
+        "tags": [NalloTags.DEEPVARIANT_REPORT],
+        "is_mandatory": True,
+        "used_by": [UsageTags.CLINICAL_DELIVERY, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["paraphase", "paraphase"]): {
+        "tags": [AlignmentTags.BAM, BioinfoToolsTags.PARAPHASE],
+        "is_mandatory": True,
+        "used_by": [UsageTags.CLINICAL_DELIVERY, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["paraphase", "paraphase_index"]): {
+        "tags": [AlignmentTags.BAM_INDEX, BioinfoToolsTags.PARAPHASE],
+        "is_mandatory": True,
+        "used_by": [UsageTags.CLINICAL_DELIVERY, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["paraphase", "json"]): {
+        "tags": [BioinfoToolsTags.PARAPHASE, ReportTags.JSON],
+        "is_mandatory": True,
+        "used_by": [UsageTags.CLINICAL_DELIVERY, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["paraphase", "vcf"]): {
+        "tags": [AlignmentTags.BAM_INDEX, BioinfoToolsTags.PARAPHASE, VariantTags.VCF],
+        "is_mandatory": False,
+        "used_by": [UsageTags.CLINICAL_DELIVERY, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["paraphase", "vcf_index"]): {
+        "tags": [AlignmentTags.BAM_INDEX, BioinfoToolsTags.PARAPHASE, VariantTags.VCF_INDEX],
+        "is_mandatory": False,
+        "used_by": [UsageTags.CLINICAL_DELIVERY, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["sorted_repeats", "vcf_str"]): {
+        "tags": [NalloTags.REPEATS, NalloTags.SORTED, VariantTags.VCF],
+        "is_mandatory": True,
+        "used_by": [UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["sorted_repeats", "vcf_str_index"]): {
+        "tags": [NalloTags.REPEATS, NalloTags.SORTED, VariantTags.VCF_INDEX],
+        "is_mandatory": True,
+        "used_by": [UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["spanning_repeats", "bam"]): {
+        "tags": [NalloTags.REPEATS, NalloTags.SPANNING, AlignmentTags.BAM],
+        "is_mandatory": True,
+        "used_by": [UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["spanning_repeats", "bam_index"]): {
+        "tags": [NalloTags.REPEATS, NalloTags.SPANNING, AlignmentTags.BAM_INDEX],
+        "is_mandatory": True,
+        "used_by": [UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["repeats_annotated", "vcf_str"]): {
+        "tags": [NalloTags.REPEATS, VariantTags.VCF_STR],
+        "is_mandatory": True,
+        "used_by": [UsageTags.SCOUT, UsageTags.CLINICAL_DELIVERY, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["repeats_annotated", "vcf_str_index"]): {
+        "tags": [NalloTags.REPEATS, NalloTags.SORTED, VariantTags.VCF_STR_INDEX],
+        "is_mandatory": True,
+        "used_by": [UsageTags.SCOUT, UsageTags.CLINICAL_DELIVERY, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["snv_annotated", "vcf_snv_research"]): {
+        "tags": [
+            AnalysisTags.RESEARCH,
+            VariantTags.VCF_SNV,
+        ],
+        "is_mandatory": True,
+        "used_by": [UsageTags.SCOUT, UsageTags.CLINICAL_DELIVERY, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["snv_annotated", "vcf_snv_research_index"]): {
+        "tags": [AnalysisTags.RESEARCH, VariantTags.VCF_SNV_INDEX],
+        "is_mandatory": True,
+        "used_by": [UsageTags.SCOUT, UsageTags.CLINICAL_DELIVERY, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["snv_annotated_filtered", "vcf_snv_clinical"]): {
+        "tags": [
+            AnalysisTags.CLINICAL,
+            VariantTags.VCF_SNV,
+        ],
+        "is_mandatory": True,
+        "used_by": [UsageTags.SCOUT, UsageTags.CLINICAL_DELIVERY, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["snv_annotated_filtered", "vcf_snv_clinical_index"]): {
+        "tags": [AnalysisTags.CLINICAL, VariantTags.VCF_SNV_INDEX],
+        "is_mandatory": True,
+        "used_by": [UsageTags.SCOUT, UsageTags.CLINICAL_DELIVERY, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["sv_annotated_ranked", "vcf_sv_research"]): {
+        "tags": [
+            AnalysisTags.RESEARCH,
+            VariantTags.VCF_SV,
+        ],
+        "is_mandatory": True,
+        "used_by": [UsageTags.SCOUT, UsageTags.CLINICAL_DELIVERY, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["sv_annotated_ranked", "vcf_sv_research_index"]): {
+        "tags": [AnalysisTags.RESEARCH, VariantTags.VCF_SV_INDEX],
+        "is_mandatory": True,
+        "used_by": [UsageTags.SCOUT, UsageTags.CLINICAL_DELIVERY, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["sv_annotated_ranked_filtered", "vcf_sv_clinical"]): {
+        "tags": [
+            AnalysisTags.CLINICAL,
+            VariantTags.VCF_SV,
+        ],
+        "is_mandatory": True,
+        "used_by": [UsageTags.SCOUT, UsageTags.CLINICAL_DELIVERY, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["sv_annotated_ranked_filtered", "vcf_sv_clinical_index"]): {
+        "tags": [AnalysisTags.CLINICAL, VariantTags.VCF_SV_INDEX],
+        "is_mandatory": True,
+        "used_by": [UsageTags.SCOUT, UsageTags.CLINICAL_DELIVERY, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["qc_bam", "mosdepth_d4"]): {
+        "tags": [AnalysisTags.COVERAGE, ReportTags.D4],
+        "is_mandatory": True,
+        "used_by": [UsageTags.SCOUT, UsageTags.CLINICAL_DELIVERY, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["copy_number", "bedgraph"]): {
+        "tags": [VariantTags.CNV, NalloTags.BEDGRAPH],
+        "is_mandatory": True,
+        "used_by": [UsageTags.CLINICAL_DELIVERY, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["depth_track", "bigwig"]): {
+        "tags": [BioinfoToolsTags.HIFICNV, AnalysisTags.BIGWIG],
+        "is_mandatory": True,
+        "used_by": [UsageTags.CLINICAL_DELIVERY, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["maf_depth_track", "bigwig"]): {
+        "tags": [BioinfoToolsTags.HIFICNV, AnalysisTags.BIGWIG, NalloTags.MAF],
+        "is_mandatory": True,
+        "used_by": [UsageTags.CLINICAL_DELIVERY, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset({"nextflow-params"}): {
+        "is_mandatory": True,
+        "tags": [NextflowTags.NEXTFLOW_PARAMS],
+        "used_by": [UsageTags.CG, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset({"nextflow-config"}): {
+        "is_mandatory": True,
+        "tags": [NextflowTags.NEXTFLOW_CONFIG],
+        "used_by": [UsageTags.CG, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset({"samplesheet"}): {
+        "is_mandatory": True,
+        "tags": [NextflowTags.SAMPLESHEET],
+        "used_by": [UsageTags.CG, UsageTags.LONG_TERM_STORAGE],
+    },
+}
+
+NALLO_TAGS = {**NALLO_COMMON_TAGS, **NEXTFLOW_TAGS}
diff --git a/cg_hermes/config/nextflow.py b/cg_hermes/config/nextflow.py
@@ -4,8 +4,6 @@
 mandatory by default. However, the tags that are available to a particular analysis is mandatory for that analysis.
 """
 
-from cg_hermes.constants.tags import UsageTags, NextflowTags
-
 from cg_hermes.constants.tags import (
     NextflowTags,
     QCTags,
@@ -26,7 +24,7 @@
         "tags": [QCTags.QC_METRICS, ReportTags.DELIVERABLE],
         "used_by": [UsageTags.CG, UsageTags.LONG_TERM_STORAGE],
     },
-    frozenset({"manifest", "manifest"}): {
+    frozenset({"manifest"}): {
         "is_mandatory": False,
         "tags": [NextflowTags.MANIFEST],
         "used_by": [UsageTags.SCOUT, UsageTags.LONG_TERM_STORAGE],