Merge 9e2a451 into 820af20

pyproject.toml

@@ -1,7 +1,7 @@
 [tool.poetry]
 # Note: Various modules refer to this system as "encoded", not "cgap-portal".
 name = "encoded"
-version = "2.4.18"
+version = "2.4.19"
 description = "Clinical Genomics Analysis Platform"
 authors = ["4DN-DCIC Team <support@4dnucleome.org>"]
 license = "MIT"
@@ -35,7 +35,7 @@ classifiers = [
 ]
 
 [tool.poetry.dependencies]
-python = ">=3.6,<3.7"
+python = ">=3.6.1,<3.7"
 boto3 = "^1.10.46"
 botocore = "^1.13.46"
 certifi = ">=2020.4.5.2"
@@ -78,7 +78,7 @@ pyramid-multiauth = "0.8.0"
 pyramid-retry = "1.0"
 pyramid-tm = "2.2.1"
 pyramid-translogger = "^0.1"
-python-dateutil = "2.5.3"
+python-dateutil = "2.7.3"
 # python-magic is presently pinned to 0.4.15 in lockstep with dcicsnovault's requirements. See explanation there.
 python-magic = "0.4.15"
 pytz = ">=2020.1"
@@ -114,6 +114,7 @@ xlwt = "1.2.0"
 "zope.interface" = "4.6.0"
 "zope.sqlalchemy" = "1.3"
 sentry-sdk = "^0.16.4"
+pandas = "^1.1.1"
 
 [tool.poetry.dev-dependencies]
 # PyCharm says boto3-stubs contains useful type hints

src/encoded/commands/create_mapping_on_deploy.py

@@ -33,6 +33,7 @@
     'FileReference',
     'Image',
     'Gene',
+    'GeneList',
     'Phenotype',
     'Disorder',
     'Individual',

src/encoded/schemas/gene_list.json

@@ -0,0 +1,59 @@
+{
+    "title": "Gene List",
+    "description": "Groups of genes that are relecant for a disease or condition",
+    "id": "/profiles/gene_list.json",
+    "$schema": "http://json-schema.org/draft-04/schema#",
+    "type": "object",
+    "required": ["institution", "project", "title"],
+    "identifyingProperties": ["uuid", "aliases"],
+    "additionalProperties": false,
+    "mixinProperties": [
+        { "$ref": "mixins.json#/schema_version" },
+        { "$ref": "mixins.json#/uuid" },
+        { "$ref": "mixins.json#/status" },
+        { "$ref": "mixins.json#/aliases" },
+        { "$ref": "mixins.json#/attribution" },
+        { "$ref": "mixins.json#/submitted" },
+        { "$ref": "mixins.json#/modified" },
+        { "$ref": "mixins.json#/tags" },
+        { "$ref": "mixins.json#/notes" }
+    ],
+    "properties": {
+        "schema_version": {
+            "default": "1"
+        },
+        "title": {
+            "title": "Title",
+            "description": "Title for this Gene List",
+            "uniqueItems": true,
+            "type": "string"
+        },
+        "genes" : {
+            "title" : "Genes",
+            "type" : "array",
+            "uniqueItems": true,
+            "items" : {
+                "title" : "Gene",
+                "type" : "string",
+                "linkTo" : "Gene"
+            }
+        },
+        "disorders" : {
+            "title" : "Related Disorders",
+            "description": "Disorders that are related to this gene list",
+            "type" : "array",
+            "uniqueItems": true,
+            "items" : {
+                "title" : "Disorder",
+                "type" : "string",
+                "linkTo" : "Disorder"
+            }
+        },
+        "source_file": {
+            "title" : "Source File",
+            "description": "The original gene list file used for generating this gene list item",
+            "type" : "string",
+            "linkTo": "Document"
+        }
+    }
+}

src/encoded/tests/data/demo_inserts/document.json

@@ -0,0 +1,29 @@
+[
+    {
+        "attachment": {
+            "download": "bgm0339-koray_V4.pbxml",
+            "type": "application/proband+xml",
+            "href": "@@download/attachment/bgm0339-koray_V4.pbxml",
+            "md5sum": "8dae78069f85fce77ea7c0a6e43e0274",
+            "blob_id": "9fdd87b7-a04d-4c93-b7e0-539527072167"
+        },
+        "date_created": "2020-07-21T17:33:45.486329+00:00",
+        "submitted_by": "986b362f-4eb6-4a9c-8173-3ab267221234",
+        "last_modified": {
+            "date_modified": "2020-07-21T18:00:23.009102+00:00",
+            "modified_by": "986b362f-4eb6-4a9c-8173-3ab267221234"
+        },
+        "institution": "828cd4fe-ebb0-4b36-ccbb-d2e3a36ca234",
+        "schema_version": "1",
+        "status": "in review",
+        "project": "91336042-f277-4f69-97e5-5c8a5f6d6058",
+        "uuid": "0938cab5-5f81-4da7-b299-46e4aa5d63f9"
+    },
+    {
+        "project": "12a92962-8265-4fc0-b2f8-cf14f05db58b",
+        "institution": "828cd4fe-ebb0-4b36-a94a-d2e3a36cc989",
+        "attachment": "gene_lists/breast_cancer.txt",
+        "status": "in review",
+        "uuid": "0938cab5-5f81-4da7-b299-46e4aad22222"
+    }
+]

src/encoded/tests/data/demo_inserts/gene_list.json

@@ -0,0 +1,15 @@
+[
+    {
+        "uuid": "117d83a2-6886-4ca0-9111-1119734cfaaa",
+        "title": "Demo Gene List - 1",
+        "project": "12a92962-8265-4fc0-b2f8-cf14f05db58b",
+        "institution": "828cd4fe-ebb0-4b36-a94a-d2e3a36cc989",
+        "genes": [
+            "b7f4b448-7961-417b-8ace-0a30ecbd9e10",
+            "b4c348af-a926-4b5d-bcad-c9d1f77b7fe0",
+            "a623d377-87c5-4c53-ac71-0b6822cd493e",
+            "a6164640-71f9-46a3-af2e-6977e747d15e",
+            "a116835b-207a-4cc8-bcb0-7ea0417c9978"
+        ]
+    }
+]

src/encoded/tests/data/documents/gene_lists/breast_cancer.txt

@@ -0,0 +1,28 @@
+ABRAXAS1 
+AKT1 
+ATM 
+BARD1 
+BRCA1 
+BRCA2 
+BRIP1 
+CDH1 
+CHEK2 
+FANCC 
+FANCM 
+MRE11 
+MUTYH 
+NBN 
+NF1 
+PALB2 
+PIK3CA 
+PTEN 
+RAD50 
+RAD51C 
+RAD51D 
+RECQL 
+RINT1 
+SDHB 
+SDHD 
+STK11 
+TP53
+XRCC2

src/encoded/tests/data/documents/gene_lists/test1.txt

@@ -0,0 +1,4 @@
+GENEID1
+GENEID2
+ENSG00000000003
+ENSG00000000004

src/encoded/tests/data/inserts/gene.json

@@ -0,0 +1,30 @@
+[
+    {
+        "project": "12a92962-8265-4fc0-b2f8-cf14f05db58b",
+        "status": "released",
+        "uuid": "cc7d83a2-6886-4ca0-9402-7c49734c1111",
+        "gene_symbol": "my_gene_1",
+        "ensgid": "ENSGmock0100001"
+    },
+    {
+        "project": "12a92962-8265-4fc0-b2f8-cf14f05db58b",
+        "status": "released",
+        "uuid": "cc7d83a2-6886-4ca0-9402-7c49734c2222",
+        "gene_symbol": "my_gene_2",
+        "ensgid": "ENSGmock0100002"
+    },
+    {
+        "project": "12a92962-8265-4fc0-b2f8-cf14f05db58b",
+        "status": "released",
+        "uuid": "cc7d83a2-6886-4ca0-9402-7c49734c3333",
+        "gene_symbol": "my_gene_3",
+        "ensgid": "ENSGmock0100003"
+    },
+    {
+        "project": "12a92962-8265-4fc0-b2f8-cf14f05db58b",
+        "status": "released",
+        "uuid": "cc7d83a2-6886-4ca0-9402-7c49734c4444",
+        "gene_symbol": "my_gene_4",
+        "ensgid": "ENSGmock0100004"
+    }
+]

src/encoded/tests/data/inserts/gene_list.json

@@ -0,0 +1,17 @@
+[
+    {
+        "project": "12a92962-8265-4fc0-b2f8-cf14f05db58b",
+        "status": "released",
+        "uuid": "cc7d83a2-6886-4ca0-9402-7c49734l1111",
+        "title": "Test Gene List for Green Thumb condition",
+        "genes": [
+            "cc7d83a2-6886-4ca0-9402-7c49734c1111",
+            "cc7d83a2-6886-4ca0-9402-7c49734c2222",
+            "cc7d83a2-6886-4ca0-9402-7c49734c3333",
+            "cc7d83a2-6886-4ca0-9402-7c49734c4444"
+        ],
+        "disorders": [
+            "71bae2e1-4532-45cf-8fd7-958b47d7e446"
+        ]
+    }
+]

src/encoded/tests/datafixtures.py

@@ -8,14 +8,14 @@
     'file_format', 'variant_consequence', 'phenotype',
     'cohort', 'family', 'individual', 'sample', 'workflow',
     'access_key', 'disorder', 'document', 'file_fastq',
-    'file_processed', 'file_reference', 'gene', 'sample_processing',
+    'file_processed', 'file_reference', 'gene', 'gene_list', 'sample_processing',
     'case', 'report', 'page', 'quality_metric_fastqc', 'evidence_dis_pheno',
     'quality_metric_bamcheck', 'quality_metric_qclist', 'quality_metric_wgs_bamqc',
     'quality_metric_cmphet', 'quality_metric_vcfcheck', 'quality_metric_workflowrun',
     'quality_metric_vcfqc', 'quality_metric_bamqc', 'quality_metric_peddyqc',
     'software', 'static_section', 'tracking_item', 'workflow_mapping',
     'workflow_run_awsem', 'workflow_run', 'annotation_field', 'variant_sample',
-    'variant', 'gene_annotation_field', 'gene',
+    'variant', 'gene_annotation_field'
 ]