Add paraphrase outputs, and change paraphase jsons to store original output per sample

Author: fellen31

cg_hermes/config/nallo.py

@@ -200,6 +200,16 @@
         "is_mandatory": True,
         "used_by": [UsageTags.CLINICAL_DELIVERY, UsageTags.LONG_TERM_STORAGE],
     },
+    frozenset(["paraphrase", "json"]): {
+        "tags": [BioinfoToolsTags.PARAPHRASE, ReportTags.JSON],
+        "is_mandatory": False,
+        "used_by": [UsageTags.CLINICAL_DELIVERY, UsageTags.LONG_TERM_STORAGE],
+    },
+    frozenset(["paraphrase", "tsv"]): {
+        "tags": [BioinfoToolsTags.PARAPHRASE, ReportTags.TSV],
+        "is_mandatory": False,
+        "used_by": [UsageTags.CLINICAL_DELIVERY, UsageTags.LONG_TERM_STORAGE],
+    },
     frozenset(["paraphase", "vcf"]): {
         "tags": [BioinfoToolsTags.PARAPHASE, VariantTags.VCF],
         "is_mandatory": False,

cg_hermes/constants/tags.py

@@ -429,6 +429,7 @@ class BioinfoToolsTags(StrEnum):
     MODKIT_PILEUP: str = "modkit-pileup"
     NEXTCLADE: str = "nextclade"
     PARAPHASE: str = "paraphase"
+    PARAPHRASE: str = "paraphrase"
     PEDDY: str = "peddy"
     PICARD: str = "picard"
     PIZZLY: str = "pizzly"
@@ -489,6 +490,7 @@ def description(self) -> str:
             self.MODKIT_PILEUP: "Modkit pileup tool output",
             self.NEXTCLADE: "Viral genome clade assignment",
             self.PARAPHASE: "Paraphase tool output",
+            self.PARAPHRASE: "Paraphrase annotated paraphase JSONs",
             self.PEDDY: "Tool to check pedigree and ancestral relations",
             self.PICARD: "Picard set of bioinformatic tools",
             self.PIZZLY: "Fusion caller",

docs/nallo_map.md

@@ -36,6 +36,8 @@
 | paraphase                                           | True        | bam, paraphase                      | clinical-delivery, long-term-storage        |
 | paraphase_index, paraphase                          | True        | bam-index, paraphase                | clinical-delivery, long-term-storage        |
 | json, paraphase                                     | True        | paraphase, json                     | clinical-delivery, long-term-storage        |
+| json, paraphrase                                     | False        | paraphrase, json                     | clinical-delivery, long-term-storage        |
+| tsv, paraphrase                                     | False        | paraphrase, tsv                     | clinical-delivery, long-term-storage        |
 | vcf, paraphase                                      | False       | paraphase, vcf                      | clinical-delivery, long-term-storage        |
 | vcf_index, paraphase                                | False       | paraphase, vcf-index                | clinical-delivery, long-term-storage        |
 | sambamba_depth                                      | True       | coverage, sambamba_depth             | cg, long-term-storage                          |

tests/fixtures/nallo/case_id_deliverables.yaml

@@ -426,10 +426,20 @@ files:
   step: paraphase
   tag: paraphase_index
 - format: json
-  id: CASEID
-  path: PATHTOCASE/paraphase/family/CASEID/CASEID_merged.json
+  id: SAMPLEID
+  path: PATHTOCASE/paraphase/sample/SAMPLEID/SAMPLEID.paraphase.json
   step: paraphase
+  tag: paraphase
+- format: json
+  id: CASEID
+  path: PATHTOCASE/paraphase/family/CASEID/CASEID_annotated.json
+  step: paraphrase
   tag: json
+- format: tsv
+  id: CASEID
+  path: PATHTOCASE/paraphase/family/CASEID/CASEID_annotated.tsv
+  step: paraphrase
+  tag: tsv
 - format: vcf
   id: CASEID
   path: PATHTOCASE/paraphase/family/CASEID/CASEID_paraphase_merged.vcf.gz