rcsb.exdb 1.21__tar.gz → 1.23__tar.gz

{rcsb_exdb-1.21 → rcsb_exdb-1.23}/HISTORY.txt

@@ -101,3 +101,5 @@
                   Update unit tests and setuptools config
  9-May-2024 V1.2  Adjust provider type exclusion input to accept a list of types
 13-May-2024 V1.21 Update requirements; fix linting
+17-Jul-2024 V1.22 Adjust LigandNeighborMappingExtractor
+20-Aug-2024 V1.23 Add support for target cofactor data loading

{rcsb_exdb-1.21/rcsb.exdb.egg-info → rcsb_exdb-1.23}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: rcsb.exdb
-Version: 1.21
+Version: 1.23
 Summary: RCSB Python ExDB data extraction and loading workflows
 Home-page: https://github.com/rcsb/py-rcsb_exdb
 Author: John Westbrook
@@ -18,19 +18,21 @@ License-File: LICENSE
 Requires-Dist: OpenEye-toolkits>=2020.2.2
 Requires-Dist: numpy
 Requires-Dist: jsonschema>=2.6.0
-Requires-Dist: rcsb.utils.io>=1.46
-Requires-Dist: rcsb.db>=1.721
+Requires-Dist: rcsb.utils.io>=1.48
+Requires-Dist: rcsb.db>=1.723
 Requires-Dist: rcsb.utils.chem>=0.79
 Requires-Dist: rcsb.utils.chemref>=0.91
 Requires-Dist: rcsb.utils.citation>=0.22
 Requires-Dist: rcsb.utils.config>=0.40
 Requires-Dist: rcsb.utils.ec>=0.25
 Requires-Dist: rcsb.utils.go>=0.18
-Requires-Dist: rcsb.utils.seq>=0.79
+Requires-Dist: rcsb.utils.seq>=0.82
+Requires-Dist: rcsb.utils.seqalign>=0.29
+Requires-Dist: rcsb.utils.targets>=0.80
 Requires-Dist: rcsb.utils.struct>=0.47
 Requires-Dist: rcsb.utils.taxonomy>=0.43
-Requires-Dist: rcsb.utils.dictionary>=1.24
-Requires-Dist: rcsb.workflow>=0.42
+Requires-Dist: rcsb.utils.dictionary>=1.26
+Requires-Dist: rcsb.workflow>=0.44
 Requires-Dist: statistics; python_version < "3.0"
 Provides-Extra: dev
 Requires-Dist: check-manifest; extra == "dev"

{rcsb_exdb-1.21 → rcsb_exdb-1.23}/rcsb/exdb/cli/ExDbExec.py

@@ -10,6 +10,7 @@
 #   9-Mar-2023 dwp Lower refChunkSize to 10 (UniProt API having trouble streaming XML responses)
 #  25-Apr-2024 dwp Add arguments and logic to support CLI usage from weekly-update workflow;
 #                  Add support for logging output to a specific file
+#  20-Aug-2024 dwp Add load_target_cofactors operation; change name of upd_targets_cofactors to upd_targets
 ##
 __docformat__ = "google en"
 __author__ = "John Westbrook"
@@ -46,7 +47,8 @@ def main():
             "upd_ref_seq",  # Update reference sequence assignments
             "upd_neighbor_interactions",
             "upd_uniprot_taxonomy",
-            "upd_targets_cofactors",
+            "upd_targets",
+            "load_target_cofactors",
             "upd_pubchem",
             "upd_entry_info",
             "upd_glycan_idx",
@@ -108,7 +110,16 @@ def main():
     exWf = ExDbWorkflow(**commonD)
     if op in ["etl_chemref", "etl_uniprot_core", "etl_tree_node_lists", "upd_ref_seq"]:
         okR = exWf.load(op, **loadD)
-    elif op in ["upd_neighbor_interactions", "upd_uniprot_taxonomy", "upd_targets_cofactors", "upd_pubchem", "upd_entry_info", "upd_glycan_idx", "upd_resource_stash"]:
+    elif op in [
+        "upd_neighbor_interactions",
+        "upd_uniprot_taxonomy",
+        "upd_targets",
+        "load_target_cofactors",
+        "upd_pubchem",
+        "upd_entry_info",
+        "upd_glycan_idx",
+        "upd_resource_stash",
+    ]:
         okR = exWf.buildExdbResource(op, **loadD)
     else:
         logger.error("Unsupported op %r", op)

{rcsb_exdb-1.21 → rcsb_exdb-1.23}/rcsb/exdb/cli/__init__.py

@@ -2,4 +2,4 @@ __docformat__ = "google en"
 __author__ = "John Westbrook"
 __email__ = "john.westbrook@rcsb.org"
 __license__ = "Apache 2.0"
-__version__ = "1.21"
+__version__ = "1.23"

{rcsb_exdb-1.21 → rcsb_exdb-1.23}/rcsb/exdb/seq/LigandNeighborMappingExtractor.py

@@ -5,6 +5,7 @@
 # Utilities to extract ligand neighbor mapping details from the exchange collections.
 #
 # Updates:
+#  17-Jul-2024 dwp Stop fetching and including rcsb_ligand_neighbors.ligand_is_bound, since no longer populating that field
 #
 ##
 __docformat__ = "google en"
@@ -58,7 +59,6 @@ class LigandNeighborMappingExtractor(object):
                     "rcsb_polymer_entity_instance_container_identifiers.entity_id",
                     "rcsb_polymer_entity_instance_container_identifiers.asym_id",
                     "rcsb_ligand_neighbors.ligand_comp_id",
-                    "rcsb_ligand_neighbors.ligand_is_bound",
                 ],
             )
             eCount = obEx.getCount()
@@ -71,8 +71,8 @@ class LigandNeighborMappingExtractor(object):
                     entityId = peiD["rcsb_polymer_entity_instance_container_identifiers"]["entity_id"]
                     ky = entryId + "_" + entityId
                     for lnD in peiD["rcsb_ligand_neighbors"] if "rcsb_ligand_neighbors" in peiD else []:
-                        if "ligand_comp_id" in lnD and "ligand_is_bound" in lnD:
-                            rD.setdefault(ky, set()).add((lnD["ligand_comp_id"], lnD["ligand_is_bound"]))
+                        if "ligand_comp_id" in lnD:
+                            rD.setdefault(ky, set()).add(lnD["ligand_comp_id"])
                         else:
                             logger.warning("%s %s missing details lnD %r", entryId, entityId, lnD)
                 except Exception as e:

{rcsb_exdb-1.21 → rcsb_exdb-1.23}/rcsb/exdb/seq/LigandNeighborMappingProvider.py

@@ -56,7 +56,7 @@ class LigandNeighborMappingProvider(StashableBase):
             list: [chem_comp_id, ... ]
         """
         try:
-            return list(set([t[0] for t in self.__mapD["mapping"][rcsbEntityId.upper()]]))
+            return list(set([t for t in self.__mapD["mapping"][rcsbEntityId.upper()]]))
         except Exception:
             return []
 

{rcsb_exdb-1.21 → rcsb_exdb-1.23}/rcsb/exdb/tests/testLigandNeighborMappingProvider.py

@@ -69,7 +69,9 @@ class LigandNeighborMappingProviderTests(unittest.TestCase):
             ok = crmP.testCache(minCount=2)
             self.assertTrue(ok)
             nL = crmP.getLigandNeighbors("3VFJ_2")
+            logger.info("ligand neighbor list nL %r", nL)
             self.assertGreaterEqual(len(nL), 4)
+            self.assertGreater(len(nL[0]), 1)
             #
         except Exception as e:
             logger.exception("Failing with %s", str(e))

{rcsb_exdb-1.21 → rcsb_exdb-1.23}/rcsb/exdb/tests/testObjectExtractor.py

@@ -102,7 +102,7 @@ class ObjectExtractorTests(unittest.TestCase):
         """Test case - extract entries subject to date restriction"""
         try:
             tU = TimeUtil()
-            tS = tU.getTimestamp(useUtc=True, before={"days": 365 * 5})
+            tS = tU.getTimestamp(useUtc=True, before={"days": 365 * 7})
             tD = tU.getDateTimeObj(tS)
             obEx = ObjectExtractor(
                 self.__cfgOb,

{rcsb_exdb-1.21 → rcsb_exdb-1.23}/rcsb/exdb/wf/ExDbWorkflow.py

@@ -8,6 +8,7 @@
 #   2-Mar-2023 dwp Add "numProc" parameter to 'upd_ref_seq' operation methods
 #   9-Mar-2023 dwp Lower refChunkSize to 10 (UniProt API having trouble streaming XML responses)
 #  25-Apr-2024 dwp Add arguments and methods to support CLI usage from weekly-update workflow
+#  20-Aug-2024 dwp Add LoadTargetCofactors step; change name of UpdateTargetsCofactors step to UpdateTargetsData
 ##
 __docformat__ = "google en"
 __author__ = "John Westbrook"
@@ -274,7 +275,16 @@ class ExDbWorkflow(object):
         logger.info("Starting operation %r\n", op)
         #
         # argument processing
-        if op not in ["upd_neighbor_interactions", "upd_uniprot_taxonomy", "upd_targets_cofactors", "upd_pubchem", "upd_entry_info", "upd_glycan_idx", "upd_resource_stash"]:
+        if op not in [
+            "upd_neighbor_interactions",
+            "upd_uniprot_taxonomy",
+            "upd_targets",
+            "load_target_cofactors",
+            "upd_pubchem",
+            "upd_entry_info",
+            "upd_glycan_idx",
+            "upd_resource_stash",
+        ]:
             logger.error("Unsupported operation %r - exiting", op)
             return False
         try:
@@ -324,8 +334,8 @@ class ExDbWorkflow(object):
                 ok = ptsWf.updateUniProtTaxonomy() and ok
                 logger.info("updateUniProtTaxonomy status %r", ok)
             #
-            elif op == "upd_targets_cofactors":
-                logger.info("Starting UpdateTargetsCofactors")
+            elif op == "upd_targets":
+                logger.info("Starting UpdateTargetsData")
                 ptsWf = ProteinTargetSequenceExecutionWorkflow(
                     configPath=self.__configPath,
                     mockTopPath=self.__mockTopPath,
@@ -352,6 +362,18 @@ class ExDbWorkflow(object):
                 logger.info("buildCofactorData status %r", ok)
                 ptsWf.resourceCheck()
             #
+            elif op == "load_target_cofactors":
+                logger.info("Starting LoadTargetCofactors")
+                ptsWf = ProteinTargetSequenceExecutionWorkflow(
+                    configPath=self.__configPath,
+                    mockTopPath=self.__mockTopPath,
+                    configName=self.__configName,
+                    cachePath=self.__cachePath,
+                )
+                ok = ptsWf.loadTargetCofactorData()
+                logger.info("loadTargetCofactorData status %r", ok)
+                ptsWf.resourceCheck()
+            #
             elif op == "upd_pubchem":
                 #  -- Update local chemical indices and  create PubChem mapping index ---
                 logger.info("Starting workflow PubChemEtlWorkflow")

{rcsb_exdb-1.21 → rcsb_exdb-1.23/rcsb.exdb.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: rcsb.exdb
-Version: 1.21
+Version: 1.23
 Summary: RCSB Python ExDB data extraction and loading workflows
 Home-page: https://github.com/rcsb/py-rcsb_exdb
 Author: John Westbrook
@@ -18,19 +18,21 @@ License-File: LICENSE
 Requires-Dist: OpenEye-toolkits>=2020.2.2
 Requires-Dist: numpy
 Requires-Dist: jsonschema>=2.6.0
-Requires-Dist: rcsb.utils.io>=1.46
-Requires-Dist: rcsb.db>=1.721
+Requires-Dist: rcsb.utils.io>=1.48
+Requires-Dist: rcsb.db>=1.723
 Requires-Dist: rcsb.utils.chem>=0.79
 Requires-Dist: rcsb.utils.chemref>=0.91
 Requires-Dist: rcsb.utils.citation>=0.22
 Requires-Dist: rcsb.utils.config>=0.40
 Requires-Dist: rcsb.utils.ec>=0.25
 Requires-Dist: rcsb.utils.go>=0.18
-Requires-Dist: rcsb.utils.seq>=0.79
+Requires-Dist: rcsb.utils.seq>=0.82
+Requires-Dist: rcsb.utils.seqalign>=0.29
+Requires-Dist: rcsb.utils.targets>=0.80
 Requires-Dist: rcsb.utils.struct>=0.47
 Requires-Dist: rcsb.utils.taxonomy>=0.43
-Requires-Dist: rcsb.utils.dictionary>=1.24
-Requires-Dist: rcsb.workflow>=0.42
+Requires-Dist: rcsb.utils.dictionary>=1.26
+Requires-Dist: rcsb.workflow>=0.44
 Requires-Dist: statistics; python_version < "3.0"
 Provides-Extra: dev
 Requires-Dist: check-manifest; extra == "dev"

{rcsb_exdb-1.21 → rcsb_exdb-1.23}/rcsb.exdb.egg-info/requires.txt

@@ -1,19 +1,21 @@
 OpenEye-toolkits>=2020.2.2
 numpy
 jsonschema>=2.6.0
-rcsb.utils.io>=1.46
-rcsb.db>=1.721
+rcsb.utils.io>=1.48
+rcsb.db>=1.723
 rcsb.utils.chem>=0.79
 rcsb.utils.chemref>=0.91
 rcsb.utils.citation>=0.22
 rcsb.utils.config>=0.40
 rcsb.utils.ec>=0.25
 rcsb.utils.go>=0.18
-rcsb.utils.seq>=0.79
+rcsb.utils.seq>=0.82
+rcsb.utils.seqalign>=0.29
+rcsb.utils.targets>=0.80
 rcsb.utils.struct>=0.47
 rcsb.utils.taxonomy>=0.43
-rcsb.utils.dictionary>=1.24
-rcsb.workflow>=0.42
+rcsb.utils.dictionary>=1.26
+rcsb.workflow>=0.44
 
 [:python_version < "3.0"]
 statistics

{rcsb_exdb-1.21 → rcsb_exdb-1.23}/requirements.txt

@@ -2,17 +2,19 @@
 OpenEye-toolkits >= 2020.2.2
 numpy
 jsonschema >= 2.6.0
-rcsb.utils.io >= 1.46
-rcsb.db >= 1.721
+rcsb.utils.io >= 1.48
+rcsb.db >= 1.723
 rcsb.utils.chem >= 0.79
 rcsb.utils.chemref >= 0.91
 rcsb.utils.citation >= 0.22
 rcsb.utils.config >= 0.40
 rcsb.utils.ec >= 0.25
 rcsb.utils.go >= 0.18
-rcsb.utils.seq >= 0.79
+rcsb.utils.seq >= 0.82
+rcsb.utils.seqalign >= 0.29
+rcsb.utils.targets >= 0.80
 rcsb.utils.struct >= 0.47
 rcsb.utils.taxonomy >= 0.43
-rcsb.utils.dictionary >= 1.24
-rcsb.workflow >= 0.42
+rcsb.utils.dictionary >= 1.26
+rcsb.workflow >= 0.44
 statistics; python_version < "3.0"