Merge pull request #79 from rcsb/dev-dwp

V1.26 Add support for accessing target cofactor data from MongoDB

HISTORY.txt

@@ -135,3 +135,4 @@
 13-May-2024 - V1.24 Update requirements
 25-Jul-2024 - V1.25 Update ligand neighbor calculations and populate them into polymer entity instance features;
                     Remove NeighborInteractionProvider from DictMethodResourceProvider, since will now be calculating interactions on the fly
+20-Aug-2024 - V1.26 Add support for accessing target cofactor data from MongoDB

azure-pipelines.yml

@@ -30,13 +30,13 @@ jobs:
   - template: azure-template-tox-job.yml
     parameters: { tox: "lint_pylint", python: "3.9", os: "linux" }
   - template: azure-template-tox-job.yml
-    parameters: { tox: "test_coverage", python: "3.9", os: "linux" }
+    parameters: { tox: "test_coverage", python: "3.9", os: "linux", fixtures: 'mongodb' }
     #
   - template: azure-template-tox-job.yml
-    parameters: { tox: "py39", python: "3.9", os: "linux" }
+    parameters: { tox: "py39", python: "3.9", os: "linux", fixtures: 'mongodb' }
   #
   - template: azure-template-tox-job.yml
-    parameters: { tox: "py39", python: "3.9", os: "macos" }
+    parameters: { tox: "py39", python: "3.9", os: "macos", fixtures: 'mongodb' }
   #
   #- template: azure-template-publish-job.yml
   #  parameters: {tox: 'py39', python: '3.9', os: 'macos'}

azure-template-tox-job.yml

@@ -56,6 +56,8 @@ jobs:
             displayName: "Install bison"
           - script: which bison
             displayName: "Check bison"
+          - script: brew install mariadb
+            displayName: "Install mariadb"
       # ----------------------------------------------
       - ${{ if startsWith(parameters.os, 'linux') }}:
           - script: lsb_release -a
@@ -110,15 +112,15 @@ jobs:
             displayName: "Start and configure mysql ..."
 
         # -----
-      - ? ${{ if and(contains(parameters.fixtures, 'mongodb'), startsWith(parameters.os, 'linux')) }}
-        : # Mongo install
+      # Mongo install
+      - ${{ if and(contains(parameters.fixtures, 'mongodb'), startsWith(parameters.os, 'linux')) }}:
           - script: |
               sudo apt-get install gnupg wget
-              wget -qO - https://www.mongodb.org/static/pgp/server-4.2.asc | sudo apt-key add -
+              wget -qO - https://www.mongodb.org/static/pgp/server-7.0.asc | sudo apt-key add -
               # sudo apt-key adv --keyserver hkp://keyserver.ubuntu.com:80 --recv 9DA31620334BD75D9DCB49F368818C72E52529D4
               sudo apt list --installed | grep mongodb
-              echo "deb [ arch=amd64,arm64 ] https://repo.mongodb.org/apt/ubuntu bionic/mongodb-org/4.2 multiverse" | sudo tee /etc/apt/sources.list.d/mongodb-org-4.2.list
-              # echo "deb [ arch=amd64,arm64 ] https://repo.mongodb.org/apt/ubuntu xenial/mongodb-org/4.2 multiverse" | sudo tee /etc/apt/sources.list.d/mongodb-org-4.2.list
+              echo "deb [ arch=amd64,arm64 ] https://repo.mongodb.org/apt/ubuntu bionic/mongodb-org/7.0 multiverse" | sudo tee /etc/apt/sources.list.d/mongodb-org-7.0.list
+              # echo "deb [ arch=amd64,arm64 ] https://repo.mongodb.org/apt/ubuntu xenial/mongodb-org/7.0 multiverse" | sudo tee /etc/apt/sources.list.d/mongodb-org-7.0.list
               sudo apt-get update
               sudo apt-get install -y mongodb-org
               sudo apt list --installed | grep mongo
@@ -129,6 +131,19 @@ jobs:
               sudo ss -tulpn
             displayName: "Start Mongo service"
         #
+      - ${{ if and(contains(parameters.fixtures, 'mongodb'), startsWith(parameters.os, 'macos')) }}:
+          - script: |
+              brew tap mongodb/brew
+              brew update
+              brew install [email protected]
+            displayName: "Installing mongodb"
+          #
+          - script: brew services start [email protected]
+            displayName: "Start Mongo service"
+          #
+          - script: brew services list
+            displayName: "Check mongodb service"
+        #
       #
       - script: 'python -c "import sys; print(sys.version); print(sys.executable)"'
         displayName: show python information

rcsb/utils/dictionary/DictMethodEntityHelper.py

@@ -17,6 +17,7 @@
 #  12-Sep-2023 dwp RO-4033: When using SIFTS alignment data, don't mix and match segments from different chains of the same entity
 #   2-Nov-2023 dwp Only populate rcsb_entity_feature_summary for features that are present
 #  18-Mar-2024 dwp Separate out gathering of entity reference sequence alignments from assignment step
+#  20-Aug-2024 dwp Add support for accessing target cofactor data from MongoDB
 ##
 """
 Helper class implements methods supporting entity-level item and category methods in the RCSB dictionary extension.
@@ -82,9 +83,9 @@ def __init__(self, **kwargs):
         self.__cardP = rP.getResource("CARDTargetAnnotationProvider instance") if rP else None
         self.__imgtP = rP.getResource("IMGTTargetFeatureProvider instance") if rP else None
         self.__sabdabP = rP.getResource("SAbDabTargetFeatureProvider instance") if rP else None
-        self.__chemblP = rP.getResource("ChEMBLTargetCofactorProvider instance") if rP else None
-        self.__dbP = rP.getResource("DrugBankTargetCofactorProvider instance") if rP else None
-        self.__phP = rP.getResource("PharosTargetCofactorProvider instance") if rP else None
+        self.__chemblA = rP.getResource("ChEMBLTargetCofactorAccessor instance") if rP else None
+        self.__dbA = rP.getResource("DrugBankTargetCofactorAccessor instance") if rP else None
+        self.__phA = rP.getResource("PharosTargetCofactorAccessor instance") if rP else None
         #
         logger.debug("Dictionary entity method helper init")
 
@@ -2040,6 +2041,9 @@ def buildRelatedTargetReferences(self, dataContainer, catName, **kwargs):
             if not dataContainer.exists("entry"):
                 return False
             #
+            if dataContainer.exists("ma_data"):
+                return False
+            #
             # Create the new target category
             if not dataContainer.exists(catName):
                 dataContainer.append(DataCategory(catName, attributeNameList=self.__dApi.getAttributeNameList(catName)))
@@ -2052,12 +2056,12 @@ def buildRelatedTargetReferences(self, dataContainer, catName, **kwargs):
             ii = cObj.getRowCount()
             #
             # --- ChEMBL
-            if self.__chemblP:
+            if self.__chemblA:
                 for entityId, eType in eTypeD.items():
                     if eType not in ["polymer", "branched"]:
                         continue
                     eId = entryId + "_" + entityId
-                    tDL = self.__chemblP.getTargets(eId)
+                    tDL = self.__chemblA.getTargets(eId)
                     dupD = {}
                     for tD in tDL:
                         if tD["query_id"] in dupD:
@@ -2083,12 +2087,12 @@ def buildRelatedTargetReferences(self, dataContainer, catName, **kwargs):
                         #
                         ii += 1
             #
-            if self.__dbP:
+            if self.__dbA:
                 for entityId, eType in eTypeD.items():
                     if eType not in ["polymer", "branched"]:
                         continue
                     eId = entryId + "_" + entityId
-                    tDL = self.__dbP.getTargets(eId)
+                    tDL = self.__dbA.getTargets(eId)
                     dupD = {}
                     for tD in tDL:
                         if tD["query_id"] in dupD:
@@ -2113,12 +2117,12 @@ def buildRelatedTargetReferences(self, dataContainer, catName, **kwargs):
                         #
                         ii += 1
             #
-            if self.__phP:
+            if self.__phA:
                 for entityId, eType in eTypeD.items():
                     if eType not in ["polymer", "branched"]:
                         continue
                     eId = entryId + "_" + entityId
-                    tDL = self.__phP.getTargets(eId)
+                    tDL = self.__phA.getTargets(eId)
                     dupD = {}
                     for tD in tDL:
                         if tD["query_id"] in dupD:
@@ -2257,6 +2261,9 @@ def buildTargetCofactors(self, dataContainer, catName, **kwargs):
             if not dataContainer.exists("entry"):
                 return False
             #
+            if dataContainer.exists("ma_data"):
+                return False
+            #
             # Create the new target category
             if not dataContainer.exists(catName):
                 dataContainer.append(DataCategory(catName, attributeNameList=self.__dApi.getAttributeNameList(catName)))
@@ -2270,12 +2277,12 @@ def buildTargetCofactors(self, dataContainer, catName, **kwargs):
             #
             #  JDW "cofactor_id" -> 'cofactor_resource_id': 'DB14031', 'target_resource_id': 'P06276',
             # --- ChEMBL
-            if self.__chemblP:
+            if self.__chemblA:
                 for entityId, eType in eTypeD.items():
                     if eType not in ["polymer", "branched"]:
                         continue
                     eId = entryId + "_" + entityId
-                    tDL = self.__chemblP.getTargets(eId)
+                    tDL = self.__chemblA.getTargets(eId)
                     # --
                     dupD = {}
                     # --
@@ -2313,12 +2320,12 @@ def buildTargetCofactors(self, dataContainer, catName, **kwargs):
                             #
                             ii += 1
             #
-            if self.__dbP:
+            if self.__dbA:
                 for entityId, eType in eTypeD.items():
                     if eType not in ["polymer", "branched"]:
                         continue
                     eId = entryId + "_" + entityId
-                    tDL = self.__dbP.getTargets(eId)
+                    tDL = self.__dbA.getTargets(eId)
                     # --
                     dupD = {}
                     # --
@@ -2348,12 +2355,12 @@ def buildTargetCofactors(self, dataContainer, catName, **kwargs):
                             #
                             ii += 1
 
-            if self.__phP:
+            if self.__phA:
                 for entityId, eType in eTypeD.items():
                     if eType not in ["polymer", "branched"]:
                         continue
                     eId = entryId + "_" + entityId
-                    tDL = self.__phP.getTargets(eId)
+                    tDL = self.__phA.getTargets(eId)
                     # --
                     dupD = {}
                     # --

rcsb/utils/dictionary/DictMethodResourceProvider.py

@@ -28,6 +28,7 @@
 #                  loading of unnecessary resources for each dataset (via providerTypeExcludeL)
 #   9-May-2024 dwp Change providerTypeExclude to be a list, 'providerTypeExcludeL'
 #  25-Jul-2024 dwp Remove NeighborInteractionProvider from provider list (since calculating on the fly from now on)
+#  20-Aug-2024 dwp Add support for accessing target cofactor data from MongoDB
 ##
 """
 Resource provider for dictionary method runner and DictMethodHelper tools.
@@ -81,10 +82,10 @@
 # ---
 from rcsb.utils.targets.CARDTargetAnnotationProvider import CARDTargetAnnotationProvider
 from rcsb.utils.targets.CARDTargetOntologyProvider import CARDTargetOntologyProvider
-from rcsb.utils.targets.ChEMBLTargetCofactorProvider import ChEMBLTargetCofactorProvider
-from rcsb.utils.targets.DrugBankTargetCofactorProvider import DrugBankTargetCofactorProvider
+from rcsb.utils.targets.ChEMBLTargetCofactorProvider import ChEMBLTargetCofactorAccessor
+from rcsb.utils.targets.DrugBankTargetCofactorProvider import DrugBankTargetCofactorAccessor
 from rcsb.utils.targets.IMGTTargetFeatureProvider import IMGTTargetFeatureProvider
-from rcsb.utils.targets.PharosTargetCofactorProvider import PharosTargetCofactorProvider
+from rcsb.utils.targets.PharosTargetCofactorProvider import PharosTargetCofactorAccessor
 from rcsb.utils.targets.SAbDabTargetFeatureProvider import SAbDabTargetFeatureProvider
 
 # --
@@ -301,24 +302,30 @@ def __init__(self, cfgOb, **kwargs):
                 "buildable": True,
                 "providerType": "optional",
             },
-            "DrugBankTargetCofactorProvider instance": {
-                "class": DrugBankTargetCofactorProvider,
-                "configArgMap": {},
-                "stashable": True,
+            "DrugBankTargetCofactorAccessor instance": {
+                "class": DrugBankTargetCofactorAccessor,
+                "configArgMap": {
+                    "cfgOb": (self.__cfgOb, "value"),
+                },
+                "stashable": False,
                 "buildable": False,
                 "providerType": "optional",
             },
-            "ChEMBLTargetCofactorProvider instance": {
-                "class": ChEMBLTargetCofactorProvider,
-                "configArgMap": {},
-                "stashable": True,
+            "ChEMBLTargetCofactorAccessor instance": {
+                "class": ChEMBLTargetCofactorAccessor,
+                "configArgMap": {
+                    "cfgOb": (self.__cfgOb, "value"),
+                },
+                "stashable": False,
                 "buildable": False,
                 "providerType": "optional",
             },
-            "PharosTargetCofactorProvider instance": {
-                "class": PharosTargetCofactorProvider,
-                "configArgMap": {},
-                "stashable": True,
+            "PharosTargetCofactorAccessor instance": {
+                "class": PharosTargetCofactorAccessor,
+                "configArgMap": {
+                    "cfgOb": (self.__cfgOb, "value"),
+                },
+                "stashable": False,
                 "buildable": False,
                 "providerType": "optional",
             },

rcsb/utils/dictionary/__init__.py

@@ -2,4 +2,4 @@
 __author__ = "John Westbrook"
 __email__ = "[email protected]"
 __license__ = "Apache 2.0"
-__version__ = "1.25"
+__version__ = "1.26"

rcsb/utils/tests-dictionary/testDictMethodResourceProvider.py

@@ -42,8 +42,9 @@ class DictmethodResourceProviderTests(unittest.TestCase):
     buildTestingCache = False
 
     def setUp(self):
-        isMac = platform.system() == "Darwin"
-        self.__excludeTypeL = None if isMac else ["optional"]
+        # isMac = platform.system() == "Darwin"
+        # self.__excludeTypeL = None if isMac else ["optional"]
+        self.__excludeTypeL = ["optional"]
         self.__mockTopPath = os.path.join(TOPDIR, "rcsb", "mock-data")
         self.__cachePath = os.path.join(HERE, "test-output", "CACHE")
         self.__configPath = os.path.join(self.__mockTopPath, "config", "dbload-setup-example.yml")

rcsb/utils/tests-dictionary/testDictMethodRunner.py

@@ -42,8 +42,10 @@
 
 class DictMethodRunnerTests(unittest.TestCase):
     def setUp(self):
-        self.__isMac = platform.system() == "Darwin"
-        self.__excludeTypeL = None if self.__isMac else ["optional"]
+        # self.__isMac = platform.system() == "Darwin"
+        # self.__excludeTypeL = None if self.__isMac else ["optional"]
+        self.__isMac = False
+        self.__excludeTypeL = ["optional"]
         self.__export = True
         self.__numProc = 2
         mockTopPath = os.path.join(TOPDIR, "rcsb", "mock-data")

rcsb/utils/tests-dictionary/testDictMethodRunnerModels.py

@@ -48,8 +48,9 @@ class DictMethodRunnerModelsTests(unittest.TestCase):
     #
     def setUp(self):
 
-        self.__isMac = platform.system() == "Darwin"
-        self.__excludeTypeL = None if self.__isMac else ["optional"]
+        # self.__isMac = platform.system() == "Darwin"
+        # self.__excludeTypeL = None if self.__isMac else ["optional"]
+        self.__excludeTypeL = ["optional"]
         self.__export = True
         self.__numProc = 2
         self.__fileLimit = 5

requirements.txt

@@ -5,12 +5,12 @@ rcsb.utils.chemref >= 0.91
 rcsb.utils.citation >= 0.22
 rcsb.utils.config >= 0.40
 rcsb.utils.ec >= 0.25
-rcsb.utils.io >= 1.46
+rcsb.utils.io >= 1.48
 rcsb.utils.multiproc >= 0.19
 rcsb.utils.repository >= 0.43
-rcsb.utils.seq >= 0.79
+rcsb.utils.seq >= 0.82
 rcsb.utils.struct >= 0.47
-rcsb.utils.targets >= 0.77
+rcsb.utils.targets >= 0.80
 rcsb.utils.taxonomy >= 0.43
 rcsb.utils.validation >= 0.31
 rcsb.utils.insilico3d >= 0.38