rcsb.exdb 1.31__py3-none-any.whl

rcsb/exdb/tests/testEntryInfoEtlWorkflow.py

@@ -0,0 +1,70 @@
+##
+# File:    EntryInfoEtlWorkflowTests.py
+# Author:  J. Westbrook
+# Date:    22-Sep-2021
+#
+# Updates:
+#
+##
+"""
+Tests for entry information extraction ETL workflow methods
+"""
+
+__docformat__ = "google en"
+__author__ = "John Westbrook"
+__email__ = "jwest@rcsb.rutgers.edu"
+__license__ = "Apache 2.0"
+
+import logging
+import os
+import platform
+import resource
+import time
+import unittest
+
+from rcsb.exdb.wf.EntryInfoEtlWorkflow import EntryInfoEtlWorkflow
+
+logging.basicConfig(level=logging.INFO, format="%(asctime)s [%(levelname)s]-%(module)s.%(funcName)s: %(message)s")
+logger = logging.getLogger()
+
+HERE = os.path.abspath(os.path.dirname(__file__))
+TOPDIR = os.path.dirname(os.path.dirname(os.path.dirname(HERE)))
+
+
+class EntryInfoEtlWorkflowTests(unittest.TestCase):
+    def setUp(self):
+        #
+        self.__cachePath = os.path.join(HERE, "test-output", "CACHE")
+        self.__configPath = os.path.join(TOPDIR, "rcsb", "mock-data", "config", "dbload-setup-example.yml")
+        self.__configName = "site_info_configuration"
+        #
+        self.__startTime = time.time()
+        logger.debug("Starting %s at %s", self.id(), time.strftime("%Y %m %d %H:%M:%S", time.localtime()))
+
+    def tearDown(self):
+        unitS = "MB" if platform.system() == "Darwin" else "GB"
+        rusageMax = resource.getrusage(resource.RUSAGE_SELF).ru_maxrss
+        logger.info("Maximum resident memory size %.4f %s", rusageMax / 10 ** 6, unitS)
+        endTime = time.time()
+        logger.info("Completed %s at %s (%.4f seconds)", self.id(), time.strftime("%Y %m %d %H:%M:%S", time.localtime()), endTime - self.__startTime)
+
+    def testUpdateEntryInfo(self):
+        """Test case - update extracted entry-level annotations."""
+        try:
+            eiP = EntryInfoEtlWorkflow(configPath=self.__configPath, configName=self.__configName, cachePath=self.__cachePath)
+            ok = eiP.update(backup=False)
+            self.assertTrue(ok)
+        except Exception as e:
+            logger.exception("Failing with %s", str(e))
+            self.fail()
+
+
+def entryInfoEtlWorkflowSuite():
+    suiteSelect = unittest.TestSuite()
+    suiteSelect.addTest(EntryInfoEtlWorkflowTests("testUpdateEntryInfo"))
+    return suiteSelect
+
+
+if __name__ == "__main__":
+    mySuite = entryInfoEtlWorkflowSuite()
+    unittest.TextTestRunner(verbosity=2).run(mySuite)

rcsb/exdb/tests/testEntryInfoProvider.py

@@ -0,0 +1,97 @@
+##
+# File:    EntryInfoProviderTests.py
+# Author:  J. Westbrook
+# Date:    22-Sep-2021
+#
+# Update:
+#
+#
+##
+"""
+Tests for accessors entry-level annotations.
+
+"""
+
+__docformat__ = "google en"
+__author__ = "John Westbrook"
+__email__ = "jwest@rcsb.rutgers.edu"
+__license__ = "Apache 2.0"
+
+import logging
+import os
+import unittest
+
+from rcsb.exdb.entry.EntryInfoProvider import EntryInfoProvider
+from rcsb.utils.config.ConfigUtil import ConfigUtil
+
+HERE = os.path.abspath(os.path.dirname(__file__))
+TOPDIR = os.path.dirname(os.path.dirname(os.path.dirname(HERE)))
+
+logging.basicConfig(level=logging.INFO, format="%(asctime)s [%(levelname)s]-%(module)s.%(funcName)s: %(message)s")
+logger = logging.getLogger()
+
+
+class EntryInfoProviderTests(unittest.TestCase):
+    doInternal = False
+
+    def setUp(self):
+        self.__cachePath = os.path.join(HERE, "test-output", "CACHE")
+        #
+        self.__mockTopPath = os.path.join(TOPDIR, "rcsb", "mock-data")
+        self.__configPath = os.path.join(TOPDIR, "rcsb", "mock-data", "config", "dbload-setup-example.yml")
+        configName = "site_info_configuration"
+        self.__cfgOb = ConfigUtil(configPath=self.__configPath, defaultSectionName=configName, mockTopPath=self.__mockTopPath)
+        #
+
+    def tearDown(self):
+        pass
+
+    def testGetEntryInfo(self):
+        minCount = 12
+        eiP = EntryInfoProvider(cachePath=self.__cachePath, useCache=False)
+        ok = eiP.testCache(minCount=0)
+        self.assertTrue(ok)
+        ok = eiP.update(self.__cfgOb, fmt="json", indent=3)
+        self.assertTrue(ok)
+        riD = eiP.getEntryInfo("4en8")
+        logger.info("riD (%d) %r", len(riD), riD)
+        rL = eiP.getEntriesByPolymerEntityCount(count=2)
+        self.assertGreaterEqual(len(rL), 5)
+        ok = eiP.testCache(minCount=minCount)
+        self.assertTrue(ok)
+        eiP = EntryInfoProvider(cachePath=self.__cachePath, useCache=True)
+        ok = eiP.testCache(minCount=minCount)
+        self.assertTrue(ok)
+
+    @unittest.skipUnless(doInternal, "Internal full test")
+    def testEntryInfoRemote(self):
+        minCount = 182000
+        configName = "site_info_remote_configuration"
+        cfgOb = ConfigUtil(configPath=self.__configPath, defaultSectionName=configName, mockTopPath=self.__mockTopPath)
+        eiP = EntryInfoProvider(cachePath=self.__cachePath, useCache=True)
+        ok = eiP.testCache(minCount=minCount)
+        ok = eiP.update(cfgOb, fmt="json", indent=3)
+        self.assertTrue(ok)
+        riD = eiP.getEntryInfo("1kip")
+        logger.info("riD (%d)", len(riD))
+        self.assertGreaterEqual(len(riD), 1)
+        configName = "site_info_configuration"
+        cfgOb = ConfigUtil(configPath=self.__configPath, defaultSectionName=configName, mockTopPath=self.__mockTopPath)
+        ok = eiP.backup(cfgOb, configName, useStash=False, useGit=True)
+        self.assertTrue(ok)
+        ok = eiP.restore(cfgOb, configName, useStash=False, useGit=True)
+        self.assertTrue(ok)
+        eiP = EntryInfoProvider(cachePath=self.__cachePath, useCache=True)
+        ok = eiP.testCache(minCount=minCount)
+
+
+def entryInfoSuite():
+    suiteSelect = unittest.TestSuite()
+    suiteSelect.addTest(EntryInfoProviderTests("testGetEntryInfo"))
+    return suiteSelect
+
+
+if __name__ == "__main__":
+
+    mySuite = entryInfoSuite()
+    unittest.TextTestRunner(verbosity=2).run(mySuite)

rcsb/exdb/tests/testGlycanEtlWorkflow.py

@@ -0,0 +1,70 @@
+##
+# File:    GlycanEtlWorkflowTests.py
+# Author:  J. Westbrook
+# Date:    30-Jun-2021
+#
+# Updates:
+#
+##
+"""
+Tests for Glycan ETL workflow methods
+"""
+
+__docformat__ = "google en"
+__author__ = "John Westbrook"
+__email__ = "jwest@rcsb.rutgers.edu"
+__license__ = "Apache 2.0"
+
+import logging
+import os
+import platform
+import resource
+import time
+import unittest
+
+from rcsb.exdb.wf.GlycanEtlWorkflow import GlycanEtlWorkflow
+
+logging.basicConfig(level=logging.INFO, format="%(asctime)s [%(levelname)s]-%(module)s.%(funcName)s: %(message)s")
+logger = logging.getLogger()
+
+HERE = os.path.abspath(os.path.dirname(__file__))
+TOPDIR = os.path.dirname(os.path.dirname(os.path.dirname(HERE)))
+
+
+class GlycanEtlWorkflowTests(unittest.TestCase):
+    def setUp(self):
+        #
+        self.__cachePath = os.path.join(HERE, "test-output", "CACHE")
+        self.__configPath = os.path.join(TOPDIR, "rcsb", "mock-data", "config", "dbload-setup-example.yml")
+        self.__configName = "site_info_configuration"
+        #
+        self.__startTime = time.time()
+        logger.debug("Starting %s at %s", self.id(), time.strftime("%Y %m %d %H:%M:%S", time.localtime()))
+
+    def tearDown(self):
+        unitS = "MB" if platform.system() == "Darwin" else "GB"
+        rusageMax = resource.getrusage(resource.RUSAGE_SELF).ru_maxrss
+        logger.info("Maximum resident memory size %.4f %s", rusageMax / 10 ** 6, unitS)
+        endTime = time.time()
+        logger.info("Completed %s at %s (%.4f seconds)", self.id(), time.strftime("%Y %m %d %H:%M:%S", time.localtime()), endTime - self.__startTime)
+
+    def testUpdateIndex(self):
+        """Test case - update the glycan corrrespondence index."""
+        try:
+            gP = GlycanEtlWorkflow(configPath=self.__configPath, configName=self.__configName, cachePath=self.__cachePath)
+            ok = gP.updateMatchedIndex(backup=False)
+            self.assertTrue(ok)
+        except Exception as e:
+            logger.exception("Failing with %s", str(e))
+            self.fail()
+
+
+def glycanEtlWorkflowSuite():
+    suiteSelect = unittest.TestSuite()
+    suiteSelect.addTest(GlycanEtlWorkflowTests("testUpdateIndex"))
+    return suiteSelect
+
+
+if __name__ == "__main__":
+    mySuite = glycanEtlWorkflowSuite()
+    unittest.TextTestRunner(verbosity=2).run(mySuite)

rcsb/exdb/tests/testGlycanProvider.py

@@ -0,0 +1,98 @@
+##
+# File:    GlycanProviderTests.py
+# Author:  J. Westbrook
+# Date:    24-May-2021
+#
+# Update:
+#
+#
+##
+"""
+Tests for accessors for managing Glycan extracted annotations.
+
+"""
+
+__docformat__ = "google en"
+__author__ = "John Westbrook"
+__email__ = "jwest@rcsb.rutgers.edu"
+__license__ = "Apache 2.0"
+
+import logging
+import os
+import unittest
+
+from rcsb.exdb.branch.GlycanProvider import GlycanProvider
+from rcsb.utils.config.ConfigUtil import ConfigUtil
+from rcsb.utils.io.MarshalUtil import MarshalUtil
+
+HERE = os.path.abspath(os.path.dirname(__file__))
+TOPDIR = os.path.dirname(os.path.dirname(os.path.dirname(HERE)))
+
+logging.basicConfig(level=logging.INFO, format="%(asctime)s [%(levelname)s]-%(module)s.%(funcName)s: %(message)s")
+logger = logging.getLogger()
+
+
+class GlycanProviderTests(unittest.TestCase):
+    def setUp(self):
+        self.__cachePath = os.path.join(HERE, "test-output", "CACHE")
+        self.__dataPath = os.path.join(HERE, "test-data")
+        self.__mU = MarshalUtil(workPath=self.__cachePath)
+        #
+        self.__mockTopPath = os.path.join(TOPDIR, "rcsb", "mock-data")
+        self.__configPath = os.path.join(TOPDIR, "rcsb", "mock-data", "config", "dbload-setup-example.yml")
+        configName = "site_info_configuration"
+        self.__cfgOb = ConfigUtil(configPath=self.__configPath, defaultSectionName=configName, mockTopPath=self.__mockTopPath)
+        #
+        self.__stashUrl = None
+        self.__stashRemotePath = os.path.join(self.__cachePath, "stash-remote")
+
+    def tearDown(self):
+        pass
+
+    def testGlycanMapping(self):
+        minCount = 12
+        gP = GlycanProvider(cachePath=self.__cachePath, useCache=False)
+        ok = gP.testCache(minCount=0)
+        self.assertTrue(ok)
+        ok = gP.update(self.__cfgOb, fmt="json", indent=3)
+        self.assertTrue(ok)
+        riD = gP.getIdentifiers()
+        logger.info("riD (%d)", len(riD))
+        ok = gP.testCache(minCount=minCount)
+        self.assertTrue(ok)
+        gP = GlycanProvider(cachePath=self.__cachePath, useCache=True)
+        ok = gP.testCache(minCount=minCount)
+        self.assertTrue(ok)
+
+    @unittest.skip("Internal test")
+    def testGlycanStashRemote(self):
+        minCount = 12
+        configName = "site_info_remote_configuration"
+        cfgOb = ConfigUtil(configPath=self.__configPath, defaultSectionName=configName, mockTopPath=self.__mockTopPath)
+        gP = GlycanProvider(cachePath=self.__cachePath, useCache=True)
+        ok = gP.testCache(minCount=minCount)
+        ok = gP.update(cfgOb, fmt="json", indent=3)
+        self.assertTrue(ok)
+        riD = gP.getIdentifiers()
+        logger.info("riD (%d)", len(riD))
+        self.assertTrue(ok)
+        ok = gP.backup(cfgOb, configName)
+        self.assertTrue(ok)
+        ok = gP.restore(cfgOb, configName)
+        self.assertTrue(ok)
+        #
+        ok = gP.update(cfgOb, fmt="json", indent=3)
+        self.assertTrue(ok)
+        #
+
+
+def glycanMappingSuite():
+    suiteSelect = unittest.TestSuite()
+    suiteSelect.addTest(GlycanProviderTests("testGlycanMapping"))
+    return suiteSelect
+
+
+if __name__ == "__main__":
+
+    mySuite = glycanMappingSuite()
+    unittest.TextTestRunner(verbosity=2).run(mySuite)

rcsb/exdb/tests/testGlycanUtils.py

@@ -0,0 +1,64 @@
+##
+# File:    GlycanUtilsTests.py
+# Author:  J. Westbrook
+# Date:    25-May-2021
+#
+# Update:
+#   9-Feb-2023 aae  Fix TOPDIR path
+##
+"""
+Tests for creating glycan accession mapping details.
+"""
+
+__docformat__ = "google en"
+__author__ = "John Westbrook"
+__email__ = "jwest@rcsb.rutgers.edu"
+__license__ = "Apache 2.0"
+
+import logging
+import os
+import unittest
+
+from rcsb.exdb.branch.GlycanUtils import GlycanUtils
+from rcsb.utils.config.ConfigUtil import ConfigUtil
+
+HERE = os.path.abspath(os.path.dirname(__file__))
+TOPDIR = os.path.dirname(os.path.dirname(os.path.dirname(HERE)))
+
+logging.basicConfig(level=logging.INFO, format="%(asctime)s [%(levelname)s]-%(module)s.%(funcName)s: %(message)s")
+logger = logging.getLogger()
+
+
+class GlycanUtilsTests(unittest.TestCase):
+    def setUp(self):
+        self.__dirPath = os.path.join(HERE, "test-output", "CACHE", "glycan")
+        #
+        self.__mockTopPath = os.path.join(TOPDIR, "rcsb", "mock-data")
+        configPath = os.path.join(TOPDIR, "rcsb", "mock-data", "config", "dbload-setup-example.yml")
+        configName = "site_info_configuration"
+        self.__cfgOb = ConfigUtil(configPath=configPath, defaultSectionName=configName, mockTopPath=self.__mockTopPath)
+        #
+
+    def tearDown(self):
+        pass
+
+    def testGlycanMapping(self):
+        gU = GlycanUtils(self.__cfgOb, self.__dirPath)
+        beD = gU.getBranchedEntityDetails()
+        self.assertGreaterEqual(len(beD), 1)
+        logger.info("branched entity descriptor details (%d)", len(beD))
+        eaD = gU.updateEntityAccessionMap()
+        logger.info("updated entity accession map length (%d)", len(eaD))
+        self.assertGreaterEqual(len(eaD), 12)
+
+
+def glycanMappingSuite():
+    suiteSelect = unittest.TestSuite()
+    suiteSelect.addTest(GlycanUtilsTests("testGlycanMapping"))
+    return suiteSelect
+
+
+if __name__ == "__main__":
+
+    mySuite = glycanMappingSuite()
+    unittest.TextTestRunner(verbosity=2).run(mySuite)

rcsb/exdb/tests/testLigandNeighborMappingProvider.py

@@ -0,0 +1,90 @@
+##
+# File:    LigandNeighborMappingProviderTests.py
+# Author:  J. Westbrook
+# Date:    28-Jun-2021
+#
+# Updates:
+#
+##
+"""
+Tests for
+
+"""
+
+__docformat__ = "google en"
+__author__ = "John Westbrook"
+__email__ = "jwest@rcsb.rutgers.edu"
+__license__ = "Apache 2.0"
+
+
+import logging
+import os
+import platform
+import resource
+import time
+import unittest
+
+from rcsb.exdb.seq.LigandNeighborMappingProvider import LigandNeighborMappingProvider
+
+from rcsb.utils.config.ConfigUtil import ConfigUtil
+
+logging.basicConfig(level=logging.INFO, format="%(asctime)s [%(levelname)s]-%(module)s.%(funcName)s: %(message)s")
+logger = logging.getLogger()
+
+HERE = os.path.abspath(os.path.dirname(__file__))
+TOPDIR = os.path.dirname(os.path.dirname(os.path.dirname(HERE)))
+
+
+class LigandNeighborMappingProviderTests(unittest.TestCase):
+    def __init__(self, methodName="runTest"):
+        super(LigandNeighborMappingProviderTests, self).__init__(methodName)
+        self.__verbose = True
+
+    def setUp(self):
+        self.__mockTopPath = os.path.join(TOPDIR, "rcsb", "mock-data")
+        configPath = os.path.join(TOPDIR, "rcsb", "mock-data", "config", "dbload-setup-example.yml")
+        configName = "site_info_configuration"
+        self.__cfgOb = ConfigUtil(configPath=configPath, defaultSectionName=configName, mockTopPath=self.__mockTopPath)
+        self.__cachePath = os.path.join(TOPDIR, "CACHE")
+        self.__startTime = time.time()
+        logger.debug("Starting %s at %s", self.id(), time.strftime("%Y %m %d %H:%M:%S", time.localtime()))
+
+    def tearDown(self):
+        unitS = "MB" if platform.system() == "Darwin" else "GB"
+        rusageMax = resource.getrusage(resource.RUSAGE_SELF).ru_maxrss
+        logger.info("Maximum resident memory size %.4f %s", rusageMax / 10 ** 6, unitS)
+        endTime = time.time()
+        logger.info("Completed %s at %s (%.4f seconds)", self.id(), time.strftime("%Y %m %d %H:%M:%S", time.localtime()), endTime - self.__startTime)
+
+    def testLigandNeighborMapping(self):
+        """Test case - load and access ligand neighbor mapping cache"""
+        try:
+            crmP = LigandNeighborMappingProvider(self.__cachePath, useCache=True)
+            ok = crmP.testCache()
+            self.assertTrue(ok)
+            #
+            ok = crmP.fetchLigandNeighborMapping(self.__cfgOb)
+            self.assertTrue(ok)
+            crmP = LigandNeighborMappingProvider(self.__cachePath, useCache=True)
+            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))
+            self.fail()
+
+
+def ligandNeighborMappingSuite():
+    suiteSelect = unittest.TestSuite()
+    suiteSelect.addTest(LigandNeighborMappingProviderTests("testLigandNeighborMapping"))
+    return suiteSelect
+
+
+if __name__ == "__main__":
+    #
+    mySuite = ligandNeighborMappingSuite()
+    unittest.TextTestRunner(verbosity=2).run(mySuite)