changes to address pr comments

Author: lindaxiang

payload-gen-variant-filtering/main.nf

@@ -49,6 +49,7 @@ params.files_to_upload = []
 params.wf_name = ""
 params.wf_short_name = ""
 params.wf_version = ""
+params.controlled = false
 
 
 process payloadGenVariantFiltering {
@@ -64,24 +65,25 @@ process payloadGenVariantFiltering {
     val wf_name
     val wf_short_name
     val wf_version
+    val controlled
 
   output:  // output, make update as needed
     path "*.payload.json", emit: payload
 
   script:
     // add and initialize variables here as needed
+    arg_controlled = controlled ? "-c" : ""
 
-    """
-    mkdir -p output_dir
-
+    """   
     main.py \
       -a ${analysis} \
       -f ${files_to_upload} \
       -w ${wf_name} \
       -s ${wf_short_name} \
       -v ${wf_version} \
       -r ${workflow.runName} \
-      -j ${workflow.sessionId}
+      -j ${workflow.sessionId} \
+      ${arg_controlled}
 
     """
 }
@@ -95,6 +97,7 @@ workflow {
     Channel.fromPath(params.files_to_upload).collect(),
     params.wf_name,
     params.wf_short_name,
-    params.wf_version
+    params.wf_version,
+    params.controlled
   )
 }

payload-gen-variant-filtering/main.py

@@ -33,6 +33,10 @@
     'indel': ['Simple Nucleotide Variation', 'Raw InDel Calls', ['Pindel', 'bcftools'], ['GATK-Mutect2', 'bcftools']]
 }
 
+workflow_full_name = {
+    'open-access-variant-filtering': 'Open Access Variant Filtering'
+}
+
 def calculate_size(file_path):
     return os.stat(file_path).st_size
 
@@ -45,16 +49,15 @@ def calculate_md5(file_path):
     return md5.hexdigest()
 
 
-def get_files_info(file_to_upload):
-    basename = os.path.basename(file_to_upload)
-    input_wf = basename.split(".")[5]
-    variant_type = basename.split(".")[7]
+def get_files_info(file_to_upload, args, input_wf, variant_type):
+    basename = os.path.basename(file_to_upload).replace("filtered", args.wf_short_name)
+    
     file_info = {
         'fileName': basename,
         'fileType': 'VCF' if basename.endswith('.vcf.gz') else basename.split(".")[-1].upper(),
         'fileSize': calculate_size(file_to_upload),
         'fileMd5sum': calculate_md5(file_to_upload),
-        'fileAccess': 'open',
+        'fileAccess': 'open' if not args.controlled else 'controlled',
         'info': {
             'data_category': variant_type_to_data_type_etc[variant_type][0]
         }
@@ -84,6 +87,16 @@ def get_sample_info(sample_list):
 
     return samples
 
+def get_variant_type(analysis):
+    for f in analysis.get('files'):
+        if f.get('dataType') == "VCF Index": continue
+        if f.get('dataType') == "Raw SNV Calls":
+            variant_type = 'snv'
+        elif f.get('dataType') == "Raw InDel Calls":
+            variant_type = 'indel'
+    
+    return variant_type
+
 def main():
     """
     Python implementation of tool: payload-gen-variant-filtering
@@ -98,13 +111,17 @@ def main():
     parser.add_argument("-v", dest="wf_version", type=str, required=True, help="workflow version")
     parser.add_argument("-r", dest="wf_run", type=str, required=True, help="workflow run ID")
     parser.add_argument("-j", dest="wf_session", type=str, required=True, help="workflow session ID")
+    parser.add_argument("-c", dest="controlled", action='store_true', help="set file to be controlled access")
     args = parser.parse_args()
 
     analysis = {}
     with open(args.analysis, 'r') as f:
         analysis = json.load(f)
 
     input_analysis_type = analysis.get('analysisType').get('name')
+    input_wf = analysis.get('workflow', {}).get('workflow_short_name')
+    variant_type = get_variant_type(analysis)
+
     output_analysis_type = "variant_filtering"
     payload = {
         'analysisType': {
@@ -115,7 +132,7 @@ def main():
         'samples': get_sample_info(analysis.get('samples')),
         'files': [],
         'workflow': {
-            'workflow_name': 'Open Access Variant Filtering',
+            'workflow_name': workflow_full_name.get(args.wf_name),
             'workflow_short_name': args.wf_short_name,
             'workflow_version': args.wf_version,
             'run_id': args.wf_run,
@@ -132,7 +149,7 @@ def main():
     }
 
     for f in args.files_to_upload:
-        file_info = get_files_info(f)
+        file_info = get_files_info(f, args, input_wf, variant_type)
         payload['files'].append(file_info)
 
     with open("%s.%s.payload.json" % (str(uuid.uuid4()), output_analysis_type), 'w') as f:

payload-gen-variant-filtering/tests/checker.nf

@@ -48,6 +48,7 @@ params.files_to_upload = []
 params.wf_name = ""
 params.wf_short_name = ""
 params.wf_version = ""
+params.controlled = false
 params.expected_output = ""
 
 include { payloadGenVariantFiltering } from '../main'
@@ -78,6 +79,7 @@ workflow checker {
     wf_name
     wf_short_name
     wf_version
+    controlled
     expected_output
 
   main:
@@ -86,7 +88,8 @@ workflow checker {
       files_to_upload,
       wf_name,
       wf_short_name,
-      wf_version
+      wf_version,
+      controlled
     )
 
     file_smart_diff(
@@ -103,6 +106,7 @@ workflow {
     params.wf_name,
     params.wf_short_name,
     params.wf_version,
+    params.controlled,
     file(params.expected_output)
   )
 }

payload-gen-variant-filtering/tests/expected/expected.gatk-mutect2.filtered.snv.payload.json

@@ -26,7 +26,7 @@
   ],
   "files": [
     {
-      "fileName": "TEST-PR.DO250183.SA610229.wgs.20200922.gatk-mutect2.somatic.snv.filtered.vcf.gz",
+      "fileName": "TEST-PR.DO250183.SA610229.wgs.20200922.gatk-mutect2.somatic.snv.open-filter.vcf.gz",
       "fileType": "VCF",
       "fileSize": 23947,
       "fileMd5sum": "1ebce71c8231aeb7ac6c06c92adfd5db",
@@ -41,7 +41,7 @@
       "dataType": "Raw SNV Calls"
     },
     {
-      "fileName": "TEST-PR.DO250183.SA610229.wgs.20200922.gatk-mutect2.somatic.snv.filtered.vcf.gz.tbi",
+      "fileName": "TEST-PR.DO250183.SA610229.wgs.20200922.gatk-mutect2.somatic.snv.open-filter.vcf.gz.tbi",
       "fileType": "TBI",
       "fileSize": 185,
       "fileMd5sum": "7a8482dc43849a4ee8bdd4c9c77fae0c",
@@ -58,10 +58,10 @@
   ],
   "workflow": {
     "workflow_name": "Open Access Variant Filtering",
-    "workflow_short_name": "variant-filtering",
+    "workflow_short_name": "open-filter",
     "workflow_version": "0.1.0",
-    "run_id": "nice_easley",
-    "session_id": "f8271e1f-a0b8-466f-9dc0-c95bd3ec9c83",
+    "run_id": "maniac_saha",
+    "session_id": "cb4d9dc6-d0b2-4998-bfae-80976e1ded55",
     "inputs": [
       {
         "input_analysis_id": "ca105e4d-d16b-461c-905e-4dd16b861cc8",

payload-gen-variant-filtering/tests/expected/expected.sanger-wgs.filtered.snv.payload.json

@@ -26,11 +26,11 @@
   ],
   "files": [
     {
-      "fileName": "TEST-PR.DO250183.SA610229.wgs.20200513.sanger-wgs.somatic.snv.filtered.vcf.gz",
+      "fileName": "TEST-PR.DO250183.SA610229.wgs.20200513.sanger-wgs.somatic.snv.open-filter.vcf.gz",
       "fileType": "VCF",
       "fileSize": 23947,
       "fileMd5sum": "1ebce71c8231aeb7ac6c06c92adfd5db",
-      "fileAccess": "open",
+      "fileAccess": "controlled",
       "info": {
         "data_category": "Simple Nucleotide Variation",
         "analysis_tools": [
@@ -41,11 +41,11 @@
       "dataType": "Raw SNV Calls"
     },
     {
-      "fileName": "TEST-PR.DO250183.SA610229.wgs.20200513.sanger-wgs.somatic.snv.filtered.vcf.gz.tbi",
+      "fileName": "TEST-PR.DO250183.SA610229.wgs.20200513.sanger-wgs.somatic.snv.open-filter.vcf.gz.tbi",
       "fileType": "TBI",
       "fileSize": 185,
       "fileMd5sum": "7a8482dc43849a4ee8bdd4c9c77fae0c",
-      "fileAccess": "open",
+      "fileAccess": "controlled",
       "info": {
         "data_category": "Simple Nucleotide Variation",
         "analysis_tools": [
@@ -58,10 +58,10 @@
   ],
   "workflow": {
     "workflow_name": "Open Access Variant Filtering",
-    "workflow_short_name": "variant-filtering",
+    "workflow_short_name": "open-filter",
     "workflow_version": "0.1.0",
-    "run_id": "crazy_elion",
-    "session_id": "fe789152-666b-4915-946b-61bacd19e292",
+    "run_id": "festering_davinci",
+    "session_id": "33388531-767b-4fd4-ba84-1c6732983ae1",
     "inputs": [
       {
         "input_analysis_id": "3414534a-47a4-492d-9453-4a47a4192d6d",

payload-gen-variant-filtering/tests/expected/expected.sanger-wxs.filtered.indel.payload.json

@@ -26,7 +26,7 @@
   ],
   "files": [
     {
-      "fileName": "TEST-PR.DO250183.SA610229.wxs.20200911.sanger-wxs.somatic.indel.filtered.vcf.gz",
+      "fileName": "TEST-PR.DO250183.SA610229.wxs.20200911.sanger-wxs.somatic.indel.open-filter.vcf.gz",
       "fileType": "VCF",
       "fileSize": 23947,
       "fileMd5sum": "1ebce71c8231aeb7ac6c06c92adfd5db",
@@ -41,7 +41,7 @@
       "dataType": "Raw InDel Calls"
     },
     {
-      "fileName": "TEST-PR.DO250183.SA610229.wxs.20200911.sanger-wxs.somatic.indel.filtered.vcf.gz.tbi",
+      "fileName": "TEST-PR.DO250183.SA610229.wxs.20200911.sanger-wxs.somatic.indel.open-filter.vcf.gz.tbi",
       "fileType": "TBI",
       "fileSize": 185,
       "fileMd5sum": "7a8482dc43849a4ee8bdd4c9c77fae0c",
@@ -58,10 +58,10 @@
   ],
   "workflow": {
     "workflow_name": "Open Access Variant Filtering",
-    "workflow_short_name": "variant-filtering",
+    "workflow_short_name": "open-filter",
     "workflow_version": "0.1.0",
-    "run_id": "exotic_booth",
-    "session_id": "154945ec-df0f-4ec8-ae23-ff8cdab626e0",
+    "run_id": "adoring_mercator",
+    "session_id": "cd1fcd50-3231-4ce6-9dec-722ed044d970",
     "inputs": [
       {
         "input_analysis_id": "775eeab5-9cbf-4443-9eea-b59cbf144389",

payload-gen-variant-filtering/tests/test-job-1.json

@@ -6,8 +6,9 @@
     ],
     "expected_output": "expected/expected.sanger-wgs.filtered.snv.payload.json",
     "wf_name": "open-access-variant-filtering",
-    "wf_short_name": "variant-filtering",
+    "wf_short_name": "open-filter",
     "wf_version": "0.1.0",
+    "controlled": true,
     "publish_dir": "outdir",
     "cpus": 1,
     "mem": 0.5

payload-gen-variant-filtering/tests/test-job-2.json

@@ -6,7 +6,7 @@
     ],
     "expected_output": "expected/expected.gatk-mutect2.filtered.snv.payload.json",
     "wf_name": "open-access-variant-filtering",
-    "wf_short_name": "variant-filtering",
+    "wf_short_name": "open-filter",
     "wf_version": "0.1.0",
     "publish_dir": "outdir",
     "cpus": 1,

payload-gen-variant-filtering/tests/test-job-3.json

@@ -6,7 +6,7 @@
     ],
     "expected_output": "expected/expected.sanger-wxs.filtered.indel.payload.json",
     "wf_name": "open-access-variant-filtering",
-    "wf_short_name": "variant-filtering",
+    "wf_short_name": "open-filter",
     "wf_version": "0.1.0",
     "publish_dir": "outdir",
     "cpus": 1,