Squashed 'modules/pmi/' changes from a4bb2ff865..4a4cd0df72

4a4cd0df72 Add method to add UniProt reference

git-subtree-dir: modules/pmi
git-subtree-split: 4a4cd0df72eea7a7bc9b478d5f6a82e12b4b89bd

modules/pmi/pyext/src/mmcif.py

@@ -38,6 +38,7 @@
 import ihm.representation
 import ihm.geometry
 import ihm.cross_linkers
+import ihm.reference
 
 
 def _assign_id(obj, seen_objs, obj_by_id):
@@ -1082,7 +1083,7 @@ def _get_alphabet(self, alphabet):
         else:
             raise TypeError("Don't know how to handle %s" % alphabet)
 
-    def add(self, component_name, sequence, offset, alphabet):
+    def add(self, component_name, sequence, offset, alphabet, uniprot):
         def entity_seq(sequence):
             # Map X to UNK
             if 'X' in sequence:
@@ -1095,7 +1096,8 @@ def entity_seq(sequence):
             d = component_name.split("@")[0].split(".")[0]
             entity = Entity(entity_seq(sequence), description=d,
                             pmi_offset=offset,
-                            alphabet=self._get_alphabet(alphabet))
+                            alphabet=self._get_alphabet(alphabet),
+                            uniprot=uniprot)
             self.system.entities.append(entity)
             self._sequence_dict[sequence] = entity
         self[component_name] = self._sequence_dict[sequence]
@@ -1198,11 +1200,18 @@ class Entity(ihm.Entity):
        removed). The actual offset (which is the integer to be added to the
        IHM numbering to get PMI numbering, or equivalently the number of
        not-represented N-terminal residues in the PMI sequence) is
-       available in the `pmi_offset` member."""
-    def __init__(self, sequence, pmi_offset, *args, **keys):
+       available in the `pmi_offset` member.
+
+       If a UniProt accession was provided for the sequence (either when
+       State.create_molecule() was called, or in the FASTA alignment file
+       header) then that is available in the `uniprot` member, and can be
+       added to the IHM system with the add_uniprot_reference method.
+       """
+    def __init__(self, sequence, pmi_offset, uniprot, *args, **keys):
         # Offset between PMI numbering and IHM; <pmi_#> = <ihm_#> + pmi_offset
         # (pmi_offset is also the number of N-terminal gaps in the FASTA file)
         self.pmi_offset = pmi_offset
+        self.uniprot = uniprot
         super().__init__(sequence, *args, **keys)
 
     def pmi_residue(self, res_id):
@@ -1214,6 +1223,24 @@ def pmi_range(self, res_id_begin, res_id_end):
         off = self.pmi_offset
         return self(res_id_begin - off, res_id_end - off)
 
+    def add_uniprot_reference(self):
+        """Add UniProt accession (if available) to the IHM system.
+           If a UniProt accession was provided for the sequence (either when
+           State.create_molecule() was called, or in the FASTA alignment file
+           header), then look this up at the UniProt web site (requires
+           network access) to get full information, and add it to the IHM
+           system. The resulting reference object is returned. If the IMP
+           and UniProt sequences are not identical, then this object may
+           need to be modified by specifying an alignment and/or
+           single-point mutations.
+        """
+        if self.uniprot:
+            print('Adding UniProt accession %s reference for entity %s'
+                  % (self.uniprot, self.description))
+            ref = ihm.reference.UniProtSequence.from_accession(self.uniprot)
+            self.references.append(ref)
+            return ref
+
 
 class AsymUnit(ihm.AsymUnit):
     """A single asymmetric unit in the system. This roughly corresponds to
@@ -1397,7 +1424,7 @@ def create_component(self, state, name, modeled, asym_name=None):
                 self.all_modeled_components.append(name)
 
     def add_component_sequence(self, state, name, seq, asym_name=None,
-                               alphabet=None):
+                               alphabet=None, uniprot=None):
         if asym_name is None:
             asym_name = name
 
@@ -1409,7 +1436,7 @@ def add_component_sequence(self, state, name, seq, asym_name=None,
             # Offset is always zero to start with; this may be modified
             # later in finalize_build() if any non-modeled N-terminal
             # residues are removed
-            self.entities.add(name, seq, 0, alphabet)
+            self.entities.add(name, seq, 0, alphabet, uniprot)
         if asym_name in self.asym_units:
             if self.asym_units[asym_name] is None:
                 # Set up a new asymmetric unit for this component

modules/pmi/pyext/src/topology/__init__.py

@@ -809,7 +809,8 @@ def _build_protocol_output(self):
                                     asym_name=self._name_with_copy)
                 po.add_component_sequence(state, name, self.sequence,
                                           asym_name=self._name_with_copy,
-                                          alphabet=self.alphabet)
+                                          alphabet=self.alphabet,
+                                          uniprot=self.uniprot)
 
     def _finalize_build(self):
         # For clones, pass the representation of the original molecule

modules/pmi/test/test_mmcif.py

@@ -315,21 +315,43 @@ def test_cif_entities(self):
         """Test _EntityMapper class"""
         system = ihm.System()
         c = IMP.pmi.mmcif._EntityMapper(system)
-        c.add('foo', 'MELS', 0, alphabet=None)
-        c.add('bar', 'SELM', 0, alphabet=IMP.pmi.alphabets.amino_acid)
-        c.add('foo_2', 'MELS', 0, alphabet=None)
+        c.add('foo', 'MELS', 0, alphabet=None, uniprot=None)
+        c.add('bar', 'SELM', 0, alphabet=IMP.pmi.alphabets.amino_acid,
+              uniprot='baracc')
+        c.add('foo_2', 'MELS', 0, alphabet=None, uniprot=None)
         self.assertRaises(TypeError, c.add, 'baz', 'MELSXX', 0,
-                          alphabet='garbage')
+                          alphabet='garbage', uniprot=None)
         self.assertEqual(len(system.entities), 2)
         self.assertIs(c['foo'], c['foo_2'])
         self.assertIsNot(c['foo'], c['bar'])
         a = system.entities
         self.assertEqual(len(a), 2)
         self.assertEqual(a[0].description, 'foo')
+        self.assertIsNone(a[0].uniprot)
         self.assertEqual(''.join(x.code for x in a[0].sequence), 'MELS')
         self.assertEqual(a[1].description, 'bar')
+        self.assertEqual(a[1].uniprot, 'baracc')
         self.assertEqual(''.join(x.code for x in a[1].sequence), 'SELM')
 
+    def test_entity_add_uniprot_reference(self):
+        """Test Entity.add_uniprot_reference()"""
+        system = ihm.System()
+        c = IMP.pmi.mmcif._EntityMapper(system)
+        c.add('foo', 'MELS', 0, alphabet=None, uniprot=None)
+        c.add('bar', 'SELM', 0, alphabet=None, uniprot='baracc')
+        # Mock out UniProtSequence.from_accession
+        orig = ihm.reference.UniProtSequence.from_accession
+        def mock_from_acc(acc):
+            return "mock+" + acc
+        try:
+            ihm.reference.UniProtSequence.from_accession = mock_from_acc
+            ref = c['foo'].add_uniprot_reference()
+            self.assertIsNone(ref)
+            ref = c['bar'].add_uniprot_reference()
+            self.assertEqual(ref, 'mock+baracc')
+        finally:
+            ihm.reference.UniProtSequence.from_accession = orig
+
     def test_all_datasets_all_group(self):
         """Test AllDatasets.get_all_group()"""
         s = ihm.System()