protein interaction queries

KarrLab · Nov 13, 2017 · 69b5a09 · 69b5a09
1 parent 83884d6
commit 69b5a09
Show file tree

Hide file tree

Showing 3 changed files with 103 additions and 13 deletions.
diff --git a/kinetic_datanator/core/data_model.py b/kinetic_datanator/core/data_model.py
@@ -239,6 +239,24 @@ class ProteinSpecie(PolymerSpecie):
     length = obj_model.core.IntegerAttribute()
     mass = obj_model.core.IntegerAttribute()
 
+class KnownProteinComplex(ProteinSpecie):
+    """ Represents a protein interaction
+
+        Attributes:
+
+    """
+    go_id = obj_model.core.StringAttribute()
+    go_dsc = obj_model.core.StringAttribute()
+    funcat_id = obj_model.core.StringAttribute()
+    funcat_dsc = obj_model.core.StringAttribute()
+    su_cmt = obj_model.core.StringAttribute()
+    complex_cmt = obj_model.core.StringAttribute()
+    disease_cmt = obj_model.core.StringAttribute()
+    class_name = obj_model.core.StringAttribute()
+    family_name = obj_model.core.StringAttribute()
+    molecular_weight = obj_model.core.FloatAttribute()
+
+
 
 class Interaction(EntityInteractionOrProperty):
     """ Represents an interaction

diff --git a/kinetic_datanator/data_query/protein_protein_interactions.py b/kinetic_datanator/data_query/protein_protein_interactions.py
@@ -1,7 +1,7 @@
 from kinetic_datanator.core import data_model
 from kinetic_datanator.core import data_query
 from kinetic_datanator.core import common_schema
-
+from sqlalchemy import or_
 class ProteintoProteinInteractionQueryGenerator(data_query.CachedDataSourceQueryGenerator):
     """ Queries Proteins to find Interactions with other Proteins """
 
@@ -30,22 +30,55 @@ def __init__(self,
     def get_observed_values(self):
         pass
 
-    def get_observed_interactions_and_complex(self,protein):
-        pass
-    # outputs datamodel interaction and complex
+    def get_observable_interactions_and_complex(self,protein):
+
+        interact = self.get_interaction_by_subunit(protein.uniprot_id).all()
+
+        interaction = []
+        index = 1
+        for item in interact:
+            resource = data_model.Resource(namespace = item._metadata.resource[0].namespace,
+                id = item._metadata.resource[0]._id)
+            interaction.append(data_model.ProteinInteraction(participant_a = item.participant_a,
+            participant_b = item.participant_b, interaction_id = item.interaction,
+            stoichiometry_a = item.stoich_a, stoichiometry_b = item.stoich_b,
+            site_a=item.site_a, site_b=item.site_b, cross_references = [resource],
+            name = 'Interaction '+ str(index)))
+            index += 1
+
+        complex_ = self.get_known_complex_by_subunit(protein.uniprot_id)
+
+        plex = []
+        for item in complex_:
+            resource = data_model.Resource(namespace = item._metadata.resource[0].namespace,
+                id = item._metadata.resource[0]._id)
+            plex.append(data_model.KnownProteinComplex(name = item.complex_name,
+            go_id = item.go_id, go_dsc = item. go_dsc, funcat_id = item.funcat_id,
+            funcat_dsc = item.funcat_dsc, su_cmt = item.su_cmt, complex_cmt = item.complex_cmt,
+            disease_cmt = item.disease_cmt, class_name = item.class_name,
+            family_name = item.family_name, molecular_weight= item.molecular_weight,
+            cross_references = [resource]))
+
+
+        ##FIXME: Put into encompassing structure
+
+        return interaction, plex
+
+
 
     def get_interaction_by_subunit(self, uniprot, select = common_schema.ProteinInteractions):
-        # q = self.data_source.session.query(select).join(common_schema.ProteinInteractions, common_schema.AbundanceData.subunit)
-        # condition = common_schema.ProteinSubunit.gene_name == gene_name
-        # return q.filter(condition)
-        pass
+        q = self.data_source.session.query(select).filter(or_(select.participant_a == 'uniprotkb:'+uniprot,
+            select.participant_b == 'uniprotkb:'+uniprot))
+        return q
+
 
     def get_known_complex_by_subunit(self, uniprot, select = common_schema.ProteinComplex):
         q = self.data_source.session.query(select).join(common_schema.ProteinSubunit, select.protein_subunit)
         condition = common_schema.ProteinSubunit.uniprot_id == uniprot
         return q.filter(condition)
 
 
-    def get_subunits_by_known_complex(self, complex, select = common_schema.ProteinSubunit):
-        q = self.data_source.session.query(select).join(common_schema.ProteinSubunit, select.protein_subunit)
-        condition = common_schema.ProteinSubunit.uniprot_id == uniprot
+    def get_subunits_by_known_complex(self, complex_name, select = common_schema.ProteinSubunit):
+        q = self.data_source.session.query(select).join(common_schema.ProteinComplex, select.proteincomplex)
+        condition = common_schema.ProteinComplex.complex_name == complex_name
+        return q.filter(condition)
diff --git a/tests/data_query/test_protein_protein_interactions.py b/tests/data_query/test_protein_protein_interactions.py
@@ -10,10 +10,42 @@ class TestProteinProteinInteractionGenerator(unittest.TestCase):
     @classmethod
     def setUpClass(self):
         self.protein_Q9CWF2 = data_model.ProteinSpecie(uniprot_id = 'Q9CWF2')
+        self.protein = data_model.ProteinSpecie(uniprot_id = 'P53622')
 
-    def test_get_interaction_by_subunit(self):
+    def get_observed_values(self):
         pass
 
+    def test_get_observable_interactions_and_complex(self):
+        q= ppi.ProteintoProteinInteractionQueryGenerator()
+
+        interaction, complex = q.get_observable_interactions_and_complex(self.protein)
+
+        self.assertEqual(set([i.participant_a for i in interaction]),
+            set([u'uniprotkb:P41810', u'uniprotkb:P41811', u'uniprotkb:P53622',
+            u'uniprotkb:P33767', u'uniprotkb:P40509']))
+        self.assertEqual(complex, [])
+
+
+    def test_get_interaction_by_subunit(self):
+        q= ppi.ProteintoProteinInteractionQueryGenerator()
+
+        ans = q.get_interaction_by_subunit('P49418').all()
+
+        self.assertEqual(set([c.site_a for c in ans]),set([u'binding-associated region:545-695(MINT-1523780)',
+        u'binding-associated region:846-854(MINT-8096500)', u'binding-associated region:626-695(MINT-8094602)',
+        u'binding-associated region:626-695(MINT-8094834)',
+        u'glutathione s tranferase tag:?-?(MINT-8096286)|binding-associated region:620-695(MINT-8096289)',
+        u'glutathione s tranferase tag:?-?(MINT-8096392)|binding-associated region:620-695(MINT-8096394)',
+        u'binding-associated region:626-695(MINT-8094740)', u'-',
+        u'binding-associated region:620-695(MINT-8095970)|glutathione s tranferase tag:?-?(MINT-8095968)',
+        u'binding-associated region:626-695(MINT-376295)', u'binding-associated region:626-695(MINT-8094630)',
+        u'binding-associated region:620-695(MINT-8096476)|glutathione s tranferase tag:?-?(MINT-8096479)',
+        u'binding-associated region:545-695(MINT-1523797)',
+        u'binding-associated region:620-695(MINT-8095931)|glutathione s tranferase tag:?-?(MINT-8095934)',
+        u'glutathione s tranferase tag:?-?(MINT-8095915)|binding-associated region:620-695(MINT-8095918)',
+        u'glutathione s tranferase tag:?-?(MINT-8096369)|binding-associated region:620-695(MINT-8096366)',
+        u'binding-associated region:626-695(MINT-8094654)', u'binding-associated region:626-695(MINT-8094709)']))
+
     def test_get_known_complex_by_subunit(self):
         q= ppi.ProteintoProteinInteractionQueryGenerator()
 
@@ -24,4 +56,11 @@ def test_get_known_complex_by_subunit(self):
         self.assertEqual(ans[0].go_id, 'GO:0006461;GO:0006897;GO:0030036')
 
     def test_get_subunits_by_known_complex(self):
-        pass
+        q= ppi.ProteintoProteinInteractionQueryGenerator()
+
+        name = 'Succinyl-CoA synthetase, GDP-forming'
+
+        ans = q.get_subunits_by_known_complex(name).all()
+
+        self.assertEqual([c.uniprot_id for c in ans],
+            [u'O19069', u'P53590', u'P53597', u'Q96I99', u'Q9WUM5', u'Q9Z2I8'])