added support for passing down "info.origin"

Author: junjun-zhang

payload-gen-variant-calling/main.py

@@ -212,13 +212,31 @@ def get_sample_info(sample_list):
 
 def main(args):
     normal_analysis = {}
+    normal_origin = None
     with open(args.normal_analysis, 'r') as f:
         normal_analysis = json.load(f)
+        if 'info' in normal_analysis and 'origin' in normal_analysis['info']:
+            normal_origin = normal_analysis['info']['origin']
 
     tumour_analysis = {}
+    tumour_origin = None
     if args.tumour_analysis:
         with open(args.tumour_analysis, 'r') as f:
             tumour_analysis = json.load(f)
+            if 'info' in tumour_analysis and 'origin' in tumour_analysis['info']:
+                tumour_origin = tumour_analysis['info']['origin']
+
+    if normal_origin is not None and tumour_origin is not None:
+        if normal_origin == tumour_origin:
+            analysis_origin = normal_origin
+        else:
+            sys.exit(f'Origins of normal and tumour analyses differ, normal: {normal_origin}, tumour: {tumour_origin}')
+    elif normal_origin is not None:
+        analysis_origin = normal_origin
+    elif tumour_origin is not None:
+        analysis_origin = tumour_origin
+    else:
+        analysis_origin = None
 
     somatic_or_germline = 'Somatic'   # default
     if args.wf_short_name in ['sanger-wgs', 'sanger-wxs', 'gatk-mutect2']:
@@ -233,6 +251,9 @@ def main(args):
         'analysisType': {
             'name': None
         },
+        'info': {
+            'origin': analysis_origin  # adding this here, so the info field will maintain it's position
+        },
         'studyId': normal_analysis.get('studyId'),  # normal/tumour analysis should always from the same study
         'experiment': {},
         'samples': [],
@@ -248,6 +269,9 @@ def main(args):
         }
     }
 
+    if analysis_origin is None:  # if origin is None, remove the info field
+        payload.pop('info')
+
     # get sample of the payload
     if somatic_or_germline == 'Somatic':  # somatic variants
         payload['samples'] = get_sample_info(tumour_analysis.get('samples'))

tests/data/1a7df8db-cb54-43e1-952f-3c458a14800a.variant_calling.payload.json

@@ -2,6 +2,9 @@
   "analysisType": {
     "name": "variant_calling"
   },
+  "info": {
+    "origin": "ICGC-25K"
+  },
   "studyId": "TEST-PR",
   "experiment": {
     "experimental_strategy": "WGS",

tests/data/66f84b1e-dad1-4981-916e-07e62ff53410.variant_calling_supplement.payload.json

@@ -2,6 +2,9 @@
   "analysisType": {
     "name": "variant_calling_supplement"
   },
+  "info": {
+    "origin": "ICGC-25K"
+  },
   "studyId": "TEST-PR",
   "experiment": {
     "experimental_strategy": "WGS",

tests/data/735a4c34-c928-4f03-957b-fe808df68f63.qc_metrics.payload.json

@@ -2,6 +2,9 @@
   "analysisType": {
     "name": "qc_metrics"
   },
+  "info": {
+    "origin": "ICGC-25K"
+  },
   "studyId": "TEST-PR",
   "experiment": {
     "experimental_strategy": "WGS",

tests/data/7ad90309-21df-4345-9903-0921df73456b.sequencing_alignment.analysis.json

@@ -1,4 +1,7 @@
 {
+  "info": {
+    "origin": "ICGC-25K"
+  },
   "analysisId": "7ad90309-21df-4345-9903-0921df73456b",
   "studyId": "TEST-PR",
   "analysisState": "PUBLISHED",

tests/data/f7ca0bd4-53a7-4663-b7b6-1b3d414e0018.variant_calling.payload.json

@@ -2,6 +2,9 @@
   "analysisType": {
     "name": "variant_calling"
   },
+  "info": {
+    "origin": "ICGC-25K"
+  },
   "studyId": "TEST-PR",
   "experiment": {
     "experimental_strategy": "WGS",