Operator to generate inputs for sample sheets

beagle_etl/fixtures/beagle_etl.operator.json

@@ -97,5 +97,16 @@
             "version": "v1.0.0",
             "slug": "AccessLegacyCNVOperator"
         }
+    },
+    {
+        "model": "beagle_etl.operator",
+        "pk": 10,
+        "fields": {
+            "active": true,
+            "recipes": "[\"None\"]",
+            "class_name": "runner.operator.access.v1_0_0.sample_sheet.AccessSampleSheetOperator",
+            "version": "v1.0.0",
+            "slug": "AccessSampleSheetOperator"
+        }
     }
 ]

fixtures/tests/10075_D.filemetadata.json

@@ -18,6 +18,7 @@
                 "runMode": "HiSeq High Output",
                 "species": "Human",
                 "sampleId": "10075_D_5",
+                "cmoSampleName": "s_C_0CREWW_L013_d",
                 "barcodeId": "DUAL_IDT_LIB_267",
                 "libraryId": "10075_D_5_1_1_1",
                 "patientId": "C-8VK0V7",
@@ -93,6 +94,7 @@
                 "runMode": "HiSeq High Output",
                 "species": "Human",
                 "sampleId": "10075_D_4",
+                "cmoSampleName": "s_C_0CREWW_L012_d",
                 "barcodeId": "DUAL_IDT_LIB_255",
                 "libraryId": "10075_D_4_1_1_1",
                 "patientId": "C-DRKHP7",
@@ -168,6 +170,7 @@
                 "runMode": "HiSeq High Output",
                 "species": "Human",
                 "sampleId": "10075_D_2",
+                "cmoSampleName": "s_C_0CREWW_L011_d",
                 "barcodeId": null,
                 "libraryId": "10075_D_2",
                 "patientId": "C-DRKHP7",
@@ -244,6 +247,7 @@
                 "runMode": "HiSeq High Output",
                 "species": "Human",
                 "sampleId": "10075_D_3",
+                "cmoSampleName": "s_C_0CREWW_L010_d",
                 "barcodeId": null,
                 "libraryId": "10075_D_3",
                 "patientId": "C-8VK0V7",
@@ -320,6 +324,7 @@
                 "runMode": "HiSeq High Output",
                 "species": "Human",
                 "sampleId": "10075_D_4",
+                "cmoSampleName": "s_C_0CREWW_L010_d",
                 "barcodeId": "DUAL_IDT_LIB_255",
                 "libraryId": null,
                 "patientId": "C-DRKHP7",
@@ -395,6 +400,7 @@
                 "runMode": "HiSeq High Output",
                 "species": "Human",
                 "sampleId": "10075_D_2",
+                "cmoSampleName": "s_C_0CREWW_L009_d",
                 "barcodeId": null,
                 "libraryId": null,
                 "patientId": "C-DRKHP7",
@@ -471,6 +477,7 @@
                 "runMode": "HiSeq High Output",
                 "species": "Human",
                 "sampleId": "10075_D_3",
+                "cmoSampleName": "s_C_0CREWW_L008_d",
                 "barcodeId": null,
                 "libraryId": null,
                 "patientId": "C-8VK0V7",
@@ -547,6 +554,7 @@
                 "runMode": "HiSeq High Output",
                 "species": "Human",
                 "sampleId": "10075_D_5",
+                "cmoSampleName": "s_C_0CREWW_L007_d",
                 "barcodeId": "DUAL_IDT_LIB_267",
                 "libraryId": null,
                 "patientId": "C-8VK0V7",
@@ -622,6 +630,7 @@
                 "runMode": "HiSeq High Output",
                 "species": "Human",
                 "sampleId": "10075_D_1",
+                "cmoSampleName": "s_C_0CREWW_L006_d",
                 "barcodeId": "DUAL_IDT_LIB_243",
                 "libraryId": "10075_D_1_1_1_1",
                 "patientId": "C-DRKHP7",
@@ -697,6 +706,7 @@
                 "runMode": "HiSeq High Output",
                 "species": "Human",
                 "sampleId": "10075_D_1",
+                "cmoSampleName": "s_C_0CREWW_L005_d",
                 "barcodeId": "DUAL_IDT_LIB_243",
                 "libraryId": null,
                 "patientId": "C-DRKHP7",

fixtures/tests/10075_D_single_TN_pair.filemetadata.json

@@ -19,6 +19,7 @@
                 "runMode": "HiSeq High Output",
                 "species": "Human",
                 "sampleId": "10075_D_5",
+                "cmoSampleName": "s_C_0CREWW_L013_d",
                 "barcodeId": "DUAL_IDT_LIB_267",
                 "libraryId": "10075_D_5_1_1_1",
                 "patientId": "C-8VK0V7",
@@ -31,6 +32,7 @@
                 "captureName": "Pool-09483_R-10075_D-Tube7_1",
                 "igocomplete": true,
                 "labHeadName": "John Smith",
+                "dnaInputNg": 12.0,
                 "barcodeIndex": "GTATTGGC-TTGTCGGT",
                 "labHeadEmail": "[email protected]",
                 "oncoTreeCode": "MEL",
@@ -95,6 +97,7 @@
                 "runMode": "HiSeq High Output",
                 "species": "Human",
                 "sampleId": "10075_D_5",
+                "cmoSampleName": "s_C_0CREWW_L012_d",
                 "barcodeId": "DUAL_IDT_LIB_267",
                 "libraryId": null,
                 "patientId": "C-8VK0V7",
@@ -107,6 +110,7 @@
                 "captureName": "Pool-09483_R-10075_D-Tube7_1",
                 "igocomplete": true,
                 "labHeadName": "John Smith",
+                "dnaInputNg": 12.0,
                 "barcodeIndex": "GTATTGGC-TTGTCGGT",
                 "labHeadEmail": "[email protected]",
                 "oncoTreeCode": "MEL",
@@ -171,6 +175,7 @@
                 "runMode": "HiSeq High Output",
                 "species": "Human",
                 "sampleId": "10075_D_3",
+                "cmoSampleName": "s_C_0CREWW_L011_d",
                 "barcodeId": null,
                 "libraryId": "10075_D_3",
                 "patientId": "C-8VK0V7",
@@ -183,6 +188,7 @@
                 "captureName": "Pool-05257_CD-06287_AY-10075_D-Tube2_1",
                 "igocomplete": true,
                 "labHeadName": "John Smith",
+                "dnaInputNg": 12.0,
                 "barcodeIndex": null,
                 "labHeadEmail": "[email protected]",
                 "oncoTreeCode": null,
@@ -248,6 +254,7 @@
                 "runMode": "HiSeq High Output",
                 "species": "Human",
                 "sampleId": "10075_D_3",
+                "cmoSampleName": "s_C_0CREWW_L010_d",
                 "barcodeId": null,
                 "libraryId": null,
                 "patientId": "C-8VK0V7",
@@ -260,6 +267,7 @@
                 "captureName": "Pool-05257_CD-06287_AY-10075_D-Tube2_1",
                 "igocomplete": true,
                 "labHeadName": "John Smith",
+                "dnaInputNg": 12.0,
                 "barcodeIndex": null,
                 "labHeadEmail": "[email protected]",
                 "oncoTreeCode": null,

runner/fixtures/runner.pipeline.json

@@ -171,5 +171,22 @@
             "operator": 9,
             "default": true
         }
+    },
+    {
+        "model": "runner.pipeline",
+        "pk": "65419097-a2b8-4d57-a8ab-c4c4cddcbffb",
+        "fields": {
+            "created_date": "2019-11-18T17:46:45.118Z",
+            "modified_date": "2019-12-05T01:12:39.854Z",
+            "name": "sample sheet",
+            "github": "[email protected]:mskcc/ACCESS-Pipeline",
+            "version": "ij/output_bam_files_instead_of_directory",
+            "entrypoint": "cwl_tools/sample_sheet/sample_sheet.cwl",
+            "output_file_group": "a975f490-1b02-4575-abae-a4f8e3667733",
+            "output_directory": "/work/access/production/runs/voyager/sample_sheets",
+            "operator": 10,
+            "default": true
+        }
     }
+
 ]

runner/operator/access/v1_0_0/sample_sheet/__init__.py

@@ -0,0 +1,64 @@
+import os
+import logging
+
+from file_system.repository.file_repository import FileRepository
+from runner.operator.operator import Operator
+from runner.serializers import APIRunCreateSerializer
+
+logger = logging.getLogger(__name__)
+
+class AccessSampleSheetOperator(Operator):
+    def get_jobs(self):
+        """
+        Convert job inputs into serialized jobs
+
+        :return: list[(serialized job info, Job)]
+        """
+        files = FileRepository.filter(queryset=self.files,
+                                      metadata={'requestId': self.request_id,
+                                                'igocomplete': True})
+
+        samples = []
+        for f in files:
+            if f.metadata["barcodeIndex"]:
+                barcodeIndex = f.metadata["barcodeIndex"].split("-")
+                index1 = barcodeIndex[0]
+
+            for lane in f.metadata["flowCellLanes"]:
+                samples.append({
+                    "Lane": lane,
+                    "SampleID": f.metadata["cmoSampleName"][2:].replace("_", "-"),
+                    "SampleRef": f.metadata["species"],
+                    "Index": index1,
+                    "Description": f.metadata["tumorOrNormal"],
+                    "Control": "N",
+                    "Recipe": f.metadata["recipe"],
+                    "Operator": "|".join(["AR", "-;-", f.metadata["sampleId"], f.metadata["sex"], "NOVASEQ"]),
+                    "SampleProject": "Project_" + self.request_id,
+                    "DnaInputNg": f.metadata["dnaInputNg"],
+                    "CaptureInputNg": f.metadata["captureInputNg"],
+                    "LibraryVolume": f.metadata["libraryVolume"],
+                    "PatientID": f.metadata["patientId"],
+                    "IgoID": f.metadata["sampleId"],
+                })
+
+        inputs = [{
+            "samples": samples
+        }]
+
+        return [
+            (
+                APIRunCreateSerializer(
+                    data={
+                        'name': "Sample Sheet: %s, %i of %i" % (self.request_id, i + 1, len(inputs)),
+                        'app': self.get_pipeline_id(),
+                        'inputs': job,
+                        'tags': {
+                            'requestId': self.request_id
+                        }
+                    }
+                ),
+                job
+             )
+            for i, job in enumerate(inputs)
+        ]

runner/tests/operator/access/sample_sheet/test_sample_sheet_operator.py

@@ -0,0 +1,36 @@
+import os
+import json
+
+from django.test import TestCase
+
+from beagle.settings import ROOT_DIR
+from beagle_etl.models import Operator
+from runner.operator.operator_factory import OperatorFactory
+from runner.operator.access.v1_0_0.sample_sheet import AccessSampleSheetOperator
+
+
+COMMON_FIXTURES = [
+    'fixtures/tests/10075_D_single_TN_pair.file.json',
+    'fixtures/tests/10075_D_single_TN_pair.filemetadata.json',
+    'runner/fixtures/runner.pipeline.json',
+    'runner/fixtures/runner.run.json',
+    'runner/fixtures/runner.operator_run.json',
+    'file_system/fixtures/file_system.filegroup.json',
+    'file_system/fixtures/file_system.filetype.json',
+    'file_system/fixtures/file_system.storage.json',
+    'beagle_etl/fixtures/beagle_etl.operator.json',
+]
+
+
+class TestSampleSheeetOperator(TestCase):
+
+    fixtures = [os.path.join(ROOT_DIR, f) for f in COMMON_FIXTURES]
+
+    def test_sample_sheet_operator(self):
+        request_id = "10075_D"
+        operator_model = Operator.objects.get(slug="AccessSampleSheetOperator")
+        operator = OperatorFactory.get_by_model(operator_model, request_id=request_id)
+        self.assertEqual(operator.get_jobs()[0][0].is_valid(), True)
+        input_json = operator.get_jobs()[0][0].initial_data['inputs']
+        #print(json.dumps(input_json))
+        self.assertEqual(input_json, {"samples": [{"Lane": 3, "SampleID": "C-0CREWW-L013-d", "SampleRef": "Human", "Index": "GTATTGGC", "Description": "Tumor", "Control": "N", "Recipe": "IMPACT468", "Operator": "AR|-;-|10075_D_5|F|NOVASEQ", "SampleProject": "Project_10075_D", "DnaInputNg": 12.0, "CaptureInputNg": "110.0", "LibraryVolume": 35.0, "PatientID": "C-8VK0V7", "IgoID": "10075_D_5"}, {"Lane": 4, "SampleID": "C-0CREWW-L013-d", "SampleRef": "Human", "Index": "GTATTGGC", "Description": "Tumor", "Control": "N", "Recipe": "IMPACT468", "Operator": "AR|-;-|10075_D_5|F|NOVASEQ", "SampleProject": "Project_10075_D", "DnaInputNg": 12.0, "CaptureInputNg": "110.0", "LibraryVolume": 35.0, "PatientID": "C-8VK0V7", "IgoID": "10075_D_5"}, {"Lane": 3, "SampleID": "C-0CREWW-L012-d", "SampleRef": "Human", "Index": "GTATTGGC", "Description": "Tumor", "Control": "N", "Recipe": "IMPACT468", "Operator": "AR|-;-|10075_D_5|F|NOVASEQ", "SampleProject": "Project_10075_D", "DnaInputNg": 12.0, "CaptureInputNg": "110.0", "LibraryVolume": 35.0, "PatientID": "C-8VK0V7", "IgoID": "10075_D_5"}, {"Lane": 4, "SampleID": "C-0CREWW-L012-d", "SampleRef": "Human", "Index": "GTATTGGC", "Description": "Tumor", "Control": "N", "Recipe": "IMPACT468", "Operator": "AR|-;-|10075_D_5|F|NOVASEQ", "SampleProject": "Project_10075_D", "DnaInputNg": 12.0, "CaptureInputNg": "110.0", "LibraryVolume": 35.0, "PatientID": "C-8VK0V7", "IgoID": "10075_D_5"}, {"Lane": 1, "SampleID": "C-0CREWW-L011-d", "SampleRef": "Human", "Index": "GTATTGGC", "Description": "Normal", "Control": "N", "Recipe": "IMPACT468", "Operator": "AR|-;-|10075_D_3|F|NOVASEQ", "SampleProject": "Project_10075_D", "DnaInputNg": 12.0, "CaptureInputNg": "110.0", "LibraryVolume": None, "PatientID": "C-8VK0V7", "IgoID": "10075_D_3"}, {"Lane": 2, "SampleID": "C-0CREWW-L011-d", "SampleRef": "Human", "Index": "GTATTGGC", "Description": "Normal", "Control": "N", "Recipe": "IMPACT468", "Operator": "AR|-;-|10075_D_3|F|NOVASEQ", "SampleProject": "Project_10075_D", "DnaInputNg": 12.0, "CaptureInputNg": "110.0", "LibraryVolume": None, "PatientID": "C-8VK0V7", "IgoID": "10075_D_3"}, {"Lane": 3, "SampleID": "C-0CREWW-L011-d", "SampleRef": "Human", "Index": "GTATTGGC", "Description": "Normal", "Control": "N", "Recipe": "IMPACT468", "Operator": "AR|-;-|10075_D_3|F|NOVASEQ", "SampleProject": "Project_10075_D", "DnaInputNg": 12.0, "CaptureInputNg": "110.0", "LibraryVolume": None, "PatientID": "C-8VK0V7", "IgoID": "10075_D_3"}, {"Lane": 1, "SampleID": "C-0CREWW-L010-d", "SampleRef": "Human", "Index": "GTATTGGC", "Description": "Normal", "Control": "N", "Recipe": "IMPACT468", "Operator": "AR|-;-|10075_D_3|F|NOVASEQ", "SampleProject": "Project_10075_D", "DnaInputNg": 12.0, "CaptureInputNg": "110.0", "LibraryVolume": None, "PatientID": "C-8VK0V7", "IgoID": "10075_D_3"}, {"Lane": 2, "SampleID": "C-0CREWW-L010-d", "SampleRef": "Human", "Index": "GTATTGGC", "Description": "Normal", "Control": "N", "Recipe": "IMPACT468", "Operator": "AR|-;-|10075_D_3|F|NOVASEQ", "SampleProject": "Project_10075_D", "DnaInputNg": 12.0, "CaptureInputNg": "110.0", "LibraryVolume": None, "PatientID": "C-8VK0V7", "IgoID": "10075_D_3"}, {"Lane": 3, "SampleID": "C-0CREWW-L010-d", "SampleRef": "Human", "Index": "GTATTGGC", "Description": "Normal", "Control": "N", "Recipe": "IMPACT468", "Operator": "AR|-;-|10075_D_3|F|NOVASEQ", "SampleProject": "Project_10075_D", "DnaInputNg": 12.0, "CaptureInputNg": "110.0", "LibraryVolume": None, "PatientID": "C-8VK0V7", "IgoID": "10075_D_3"}]})