Metadata schema updates (#140)

* Change process/protocol_type to type

* Change file_format to format

* Fix unit tests

* Update pipeline-tools version

Author: samanehsan

adapter_pipelines/Optimus/adapter.wdl

@@ -139,7 +139,7 @@ workflow AdapterOptimus {
   Int max_cromwell_retries = 0
   Boolean add_md5s = false
 
-  String pipeline_tools_version = "v0.48.3"
+  String pipeline_tools_version = "v0.49.0"
 
   call GetInputs as prep {
     input:

adapter_pipelines/cellranger/adapter.wdl

@@ -150,7 +150,7 @@ workflow Adapter10xCount {
   Int max_cromwell_retries = 0
   Boolean add_md5s = false
 
-  String pipeline_tools_version = "v0.48.3"
+  String pipeline_tools_version = "v0.49.0"
 
   call GetInputs {
     input:

adapter_pipelines/ss2_single_sample/adapter.wdl

@@ -82,7 +82,7 @@ workflow AdapterSmartSeq2SingleCell{
   Int max_cromwell_retries = 0
   Boolean add_md5s = false
 
-  String pipeline_tools_version = "v0.48.3"
+  String pipeline_tools_version = "v0.49.0"
 
   call GetInputs as prep {
     input:

adapter_pipelines/ss2_single_sample/adapter_example_static.json

@@ -13,9 +13,9 @@
   "AdapterSmartSeq2SingleCell.reference_bundle": "bf51d668-3e14-4843-9bc7-5d676fdf0e01",
   "AdapterSmartSeq2SingleCell.format_map": "gs://hca-dcp-mint-test-data/adapters/file_format_map.json",
   "AdapterSmartSeq2SingleCell.method": "SmartSeq2SingleCell",
-  "AdapterSmartSeq2SingleCell.analysis_file_version": "5.3.6",
-  "AdapterSmartSeq2SingleCell.analysis_protocol_schema_version": "8.0.7",
-  "AdapterSmartSeq2SingleCell.analysis_process_schema_version": "9.0.0",
+  "AdapterSmartSeq2SingleCell.analysis_file_version": "6.0.0",
+  "AdapterSmartSeq2SingleCell.analysis_protocol_schema_version": "9.0.0",
+  "AdapterSmartSeq2SingleCell.analysis_process_schema_version": "11.0.1",
   "AdapterSmartSeq2SingleCell.run_type": "run",
   "AdapterSmartSeq2SingleCell.add_md5s": false
 }

pipeline_tools/create_analysis_metadata.py

@@ -61,7 +61,7 @@ def create_analysis_process(
         ),
         'schema_type': SCHEMA_TYPE,
         'process_core': get_analysis_process_core(analysis_workflow_id=analysis_id),
-        'process_type': get_analysis_process_type(),
+        'type': get_analysis_process_type(),
         'timestamp_start_utc': format_timestamp(workflow_metadata.get('start')),
         'timestamp_stop_utc': format_timestamp(workflow_metadata.get('end')),
         'input_bundles': input_bundles_string.split(','),
@@ -110,7 +110,7 @@ def create_analysis_protocol(
         'schema_type': SCHEMA_TYPE,
         'protocol_core': get_analysis_protocol_core(pipeline_version=pipeline_version),
         'computational_method': method,
-        'protocol_type': get_analysis_protocol_type(),
+        'type': get_analysis_protocol_type(),
     }
     return analysis_protocol
 
@@ -161,7 +161,7 @@ def get_outputs(output_urls, extension_to_format, schema_url, analysis_file_vers
             'schema_type': 'file',
             'file_core': {
                 'file_name': output_url.split('/')[-1],
-                'file_format': get_file_format(output_url, extension_to_format),
+                'format': get_file_format(output_url, extension_to_format),
             },
         }
         for output_url in sorted(output_urls)

pipeline_tools/tests/test_create_analysis_metadata.py

@@ -20,7 +20,7 @@ class Data:
                 'schema_type': 'file',
                 'file_core': {
                     'file_name': 'Aligned.sortedByCoord.out.bam',
-                    'file_format': 'bam',
+                    'format': 'bam',
                     'checksum': '0123456789abcdef0123456789abcdef',
                 },
             },
@@ -29,7 +29,7 @@ class Data:
                 'schema_type': 'file',
                 'file_core': {
                     'file_name': 'GSM1957573_rna_metrics',
-                    'file_format': 'metrics',
+                    'format': 'metrics',
                     'checksum': 'abcdef0123456789abcdef0123456789',
                 },
             },
@@ -257,8 +257,8 @@ def verify_outputs(
         assert output_json[0]['describedBy'] == schema_url
         assert output_json[0]['schema_type'] == 'file'
         assert (
-            output_json[0]['file_core']['file_format']
-            == expected_outputs[0]['file_core']['file_format']
+            output_json[0]['file_core']['format']
+            == expected_outputs[0]['file_core']['format']
         )
         assert (
             output_json[0]['file_core']['file_name']
@@ -272,8 +272,8 @@ def verify_outputs(
         assert output_json[0]['describedBy'] == schema_url
         assert output_json[1]['schema_type'] == 'file'
         assert (
-            output_json[1]['file_core']['file_format']
-            == expected_outputs[1]['file_core']['file_format']
+            output_json[1]['file_core']['format']
+            == expected_outputs[1]['file_core']['format']
         )
         assert (
             output_json[1]['file_core']['file_name']

pipeline_tools/tests/test_create_envelope.py

@@ -443,7 +443,7 @@ def _request_callback(request, context):
             'schema_type': 'file',
             'file_core': {
                 'file_name': 'test',
-                'file_format': 'bam',
+                'format': 'bam',
                 'checksum': '0123456789abcdef0123456789abcdef',
             },
         }
@@ -467,7 +467,7 @@ def _request_callback(request, context):
             'schema_type': 'file',
             'file_core': {
                 'file_name': 'test',
-                'file_format': 'bam',
+                'format': 'bam',
                 'checksum': '0123456789abcdef0123456789abcdef',
             },
         }
@@ -492,7 +492,7 @@ def _request_callback(request, context):
             'schema_type': 'file',
             'file_core': {
                 'file_name': 'test',
-                'file_format': 'bam',
+                'format': 'bam',
                 'checksum': '0123456789abcdef0123456789abcdef',
             },
         }