Update testing

Author: cthoyt

src/bio2bel_hgnc/__init__.py

@@ -1,6 +1,6 @@
 # -*- coding: utf-8 -*-
 
-"""Exports HGNC Equivalences"""
+"""A package for converting HGNC to BEL."""
 
 from .manager import Manager
 

src/bio2bel_hgnc/constants.py

@@ -7,7 +7,6 @@
 MODULE_NAME = 'hgnc'
 #: The default directory where PyBEL files, including logs and the  default cache, are stored. Created if not exists.
 DATA_DIR = get_data_dir(MODULE_NAME)
-DEFAULT_CACHE_CONNECTION = get_connection(MODULE_NAME)
 
 GENE_FAMILY_KEYWORD = 'GFAM'
 

src/bio2bel_hgnc/enrich.py

@@ -8,7 +8,6 @@
 
 __all__ = [
     'get_node',
-    'add_metadata',
     'add_node_orthologies',
     'add_orthologies',
     'add_node_central_dogma',
@@ -28,16 +27,6 @@ def get_node(graph, node, connection=None):
     return manager.get_node(graph, node)
 
 
-def add_metadata(graph, node, manager=None):
-    """Add to the node based on PyHGNC's database
-
-    :param pybel.BELGraph graph: A BEL Graph
-    :param tuple node: A PyBEL node tuple
-    :param Optional[str or bio2bel_hgnc.Manager] manager: A database manager
-    """
-    raise NotImplementedError
-
-
 def add_node_orthologies(graph, node, manager=None, add_leaves=False):
     """Given a node that's HGNC, add orthology relationships
 

src/bio2bel_hgnc/manager.py

@@ -2,20 +2,24 @@
 
 """Bio2BEL HGNC Manager."""
 
+import logging
 from collections import Counter
 
 import click
-import logging
-from tqdm import tqdm
-
-from bio2bel.namespace_manager import NamespaceManagerMixin
 from pybel import BELGraph
 from pybel.constants import FUNCTION, GENE, IDENTIFIER, NAME, NAMESPACE, NAMESPACE_DOMAIN_GENE, PROTEIN, RNA
 from pybel.dsl import gene as gene_dsl, protein as protein_dsl, rna as rna_dsl
 from pybel.manager.models import NamespaceEntry
+from pybel.resources import write_namespace
+from tqdm import tqdm
+
+from bio2bel.namespace_manager import NamespaceManagerMixin
 from .constants import GENE_FAMILY_KEYWORD, MODULE_NAME, encodings
 from .gfam_manager import Manager as GfamManager
-from .model_utils import *
+from .model_utils import (
+    family_to_bel, gene_to_bel, gene_to_mirna_to_bel, gene_to_protein_to_bel, gene_to_rna_to_bel,
+    uniprot_to_pybel,
+)
 from .models import Base, GeneFamily, HumanGene, MouseGene, RatGene, UniProt
 from .wrapper import BaseManager
 
@@ -51,7 +55,6 @@ def _write_gene_bel_namespace_helper(values, file):
     :param dict[str,str] values:
     :param file:
     """
-    from pybel.resources import write_namespace
     write_namespace(
         namespace_name='HGNC Human Gene Names',
         namespace_keyword='HGNC',
@@ -69,7 +72,6 @@ def _write_gene_families_bel_namespace_helper(values, file):
     :param list[str] values:
     :param file:
     """
-    from pybel.resources import write_namespace
     write_namespace(
         namespace_name='HGNC Gene Families',
         namespace_keyword='GFAM',
@@ -196,15 +198,6 @@ def get_gene_by_mgi_id(self, mgi_id):
 
         return human_genes[0]
 
-    def get_gene_by_mgi_symbol(self, mgi_symbol):
-        """Gets a HGNC gene by an orthologous MGI gene symbol
-
-        :param str mgi_symbol: MGI gene symbol
-        :rtype: Optional[bio2bel_hgnc.models.HumanGene]
-        """
-        # TODO this data is not in HGNC...
-        raise NotImplementedError
-
     def get_gene_by_rgd_id(self, rgd_id):
         """Gets a HGNC gene by an orthologous RGD identifier
 
@@ -225,15 +218,6 @@ def get_gene_by_rgd_id(self, rgd_id):
 
         return human_genes[0]
 
-    def get_gene_by_rgd_symbol(self, rgd_symbol):
-        """Gets a HGNC gene by an orthologous RGD identifier
-
-        :param str rgd_symbol: RGD gene symbol
-        :rtype: Optional[bio2bel_hgnc.models.HumanGene]
-        """
-        # TODO this data is not in HGNC...
-        raise NotImplementedError
-
     def get_node(self, graph, node):
         """Gets a node from the database, whether it has a HGNC, RGD, MGI, or EG identifier.
 
@@ -591,20 +575,33 @@ def get_all_hgnc_symbols_family(self):
         return set(res)
 
     def add_central_dogma(self, graph, node):
-        """
+        """Add the central dogma of biology.
 
         :param graph:
         :param node:
-        :return:
         """
         if node not in graph:
             raise ValueError
 
         human_gene = self.get_node(graph, node)
         encoding = encodings.get(human_gene.locus_type, 'GRP')
 
-        if 'R' in encoding:
-            graph.add_unqualified_edge(node, )
+        if 'M' in encoding:
+            rna = gene_to_rna_to_bel(human_gene)
+            graph.add_transcription(gene_to_bel(human_gene), rna)
+        elif 'R' in encoding:
+            rna = gene_to_mirna_to_bel(human_gene)
+            graph.add_transcription(gene_to_bel(human_gene), rna)
+        else:
+            return
+
+        if 'P' not in encoding:
+            return rna
+        else:
+            protein = gene_to_protein_to_bel(human_gene)
+            graph.add_translation(rna, protein)
+            return protein
+
 
     def add_node_equivalencies(self, graph, node, add_leaves=True):
         """Given an HGNC node, add equivalencies found in the database.

src/bio2bel_hgnc/model_utils.py

@@ -2,11 +2,12 @@
 
 """Bio2BEL HGNC Model utilities."""
 
-from pybel.dsl import gene as gene_dsl, protein as protein_dsl, rna as rna_dsl
+from pybel.dsl import gene as gene_dsl, mirna as mirna_dsl, protein as protein_dsl, rna as rna_dsl
 
 __all__ = [
     'gene_to_bel',
     'gene_to_rna_to_bel',
+    'gene_to_mirna_to_bel',
     'gene_to_protein_to_bel',
     'family_to_bel',
     'uniprot_to_pybel',
@@ -39,6 +40,19 @@ def gene_to_rna_to_bel(gene):
     )
 
 
+def gene_to_mirna_to_bel(gene):
+    """Converts a Gene to a PyBEL miRNA
+
+    :param bio2bel_hgnc.models.HumanGene gene:  A Gene model
+    :rtype: pybel.dsl.mirna
+    """
+    return mirna_dsl(
+        namespace='HGNC',
+        name=str(gene.symbol),
+        identifier=str(gene.identifier)
+    )
+
+
 def gene_to_protein_to_bel(gene):
     """Converts a Gene to a PyBEL gene
 

src/bio2bel_hgnc/wrapper.py

@@ -1,8 +1,10 @@
 # -*- coding: utf-8 -*-
 
+"""Wrap PyHGNC functionality."""
+
 from pyhgnc.manager.database import DbManager
 from pyhgnc.manager.query import QueryManager
 
 
 class BaseManager(DbManager, QueryManager):
-    """Wraps the relevant PyHGNC functions"""
+    """Wraps the PyHGNC managers."""

tests/__init__.py

@@ -0,0 +1 @@
+# -*- coding: utf-8 -*-

tests/cases.py

@@ -0,0 +1,18 @@
+# -*- coding: utf-8 -*-
+
+"""Test cases for Bio2BEL HGNC."""
+
+from bio2bel.testing import AbstractTemporaryCacheClassMixin
+from bio2bel_hgnc import Manager
+from tests.constants import hcop_test_path, hgnc_test_path
+
+
+class TemporaryCacheMixin(AbstractTemporaryCacheClassMixin):
+    Manager = Manager
+
+    @classmethod
+    def populate(cls):
+        cls.manager.populate(
+            hgnc_file_path=hgnc_test_path,
+            hcop_file_path=hcop_test_path,
+        )

tests/constants.py

@@ -1,5 +1,7 @@
 # -*- coding: utf-8 -*-
 
+"""Test constants for Bio2BEL HGNC."""
+
 import logging
 import os
 

tests/test_enrich_metadata.py

@@ -1,21 +1,18 @@
 # -*- coding: utf-8 -*-
 
 import logging
-import os
-import tempfile
 import unittest
 
-import pyhgnc
-from bio2bel_hgnc import Manager
-from bio2bel_hgnc.constants import GENE_FAMILY_KEYWORD
-from bio2bel_hgnc.enrich import (
-    add_metadata, add_node_central_dogma, add_node_equivalencies, add_node_orthologies, get_node,
-)
 from pybel import BELGraph
 from pybel.constants import EQUIVALENT_TO, ORTHOLOGOUS, RELATION, TRANSCRIBED_TO, TRANSLATED_TO, unqualified_edge_code
 from pybel.dsl import gene, mirna, protein, rna
 from pybel.tokens import node_to_tuple
-from tests.constants import hcop_test_path, hgnc_test_path
+
+from bio2bel_hgnc.constants import GENE_FAMILY_KEYWORD
+from bio2bel_hgnc.enrich import (
+    add_node_equivalencies, add_node_orthologies, get_node,
+)
+from tests.cases import TemporaryCacheMixin
 
 log = logging.getLogger(__name__)
 
@@ -24,37 +21,6 @@
 equivalence_code = unqualified_edge_code[EQUIVALENT_TO]
 
 
-class TemporaryCacheMixin(unittest.TestCase):
-    """
-    :type file_handle: int
-    :type path: str
-    :type manager: pyhgnc.manager.query.QueryManager
-    """
-    file_handle, path, manager = None, None, None
-
-    @classmethod
-    def setUpClass(cls):
-        """Create and populate temporary PyHGNC cache"""
-        cls.file_handle, cls.path = tempfile.mkstemp()
-        cls.connection = 'sqlite:///' + cls.path
-        log.info('Test generated connection string %s', cls.connection)
-
-        pyhgnc.update(
-            connection=cls.connection,
-            hgnc_file_path=hgnc_test_path,
-            hcop_file_path=hcop_test_path,
-        )
-
-        cls.manager = Manager(connection=cls.connection)
-
-    @classmethod
-    def tearDownClass(cls):
-        """Deletes the temporary PyHGNC cache"""
-        cls.manager.session.close()
-        os.close(cls.file_handle)
-        os.remove(cls.path)
-
-
 class TestEnrich(TemporaryCacheMixin):
     def help_check_cd33_model(self, model):
         """Checks if the given model is CD33
@@ -138,22 +104,6 @@ def test_get_entrez_node(self):
         cd33_model = get_node(graph, cd33_tuple, connection=self.manager)
         self.help_check_cd33_model(cd33_model)
 
-    def test_add_metadata(self):
-        graph = BELGraph()
-
-        cd33_test = protein(namespace='HGNC', name='CD33')
-        graph.add_node_from_data(cd33_test)
-
-        self.assertIn(cd33_test.as_tuple(), graph, msg='Graph is missing CD33 protein node')
-        self.assertIsNone(graph.get_node_label(cd33_test.as_tuple()), msg='CD33 should not have label information')
-        self.assertIsNone(graph.get_node_identifier(cd33_test.as_tuple()), msg='CD33 should not have identifier information')
-
-        add_metadata(graph, cd33_test.as_tuple(), manager=self.manager)
-
-        self.assertIn(cd33_test.as_tuple(), graph, msg='Graph somehow lost CD33 protein node')
-
-        self.assertEqual('1659', graph.get_node_identifier(cd33_test.as_tuple()), msg='Graph should be enriched with identifier')
-
     def test_add_equivalency(self):
         graph = BELGraph()
         cd33_hgnc_tuple = graph.add_node_from_data(protein(name='CD33', namespace='HGNC'))
@@ -162,7 +112,7 @@ def test_add_equivalency(self):
         self.assertEqual(2, graph.number_of_nodes())
         self.assertEqual(0, graph.number_of_edges())
 
-        add_node_equivalencies(graph, cd33_hgnc_tuple, connection=self.manager, add_leaves=False)
+        self.manager.add_node_equivalencies(graph, cd33_hgnc_tuple, add_leaves=False)
 
         self.assertEqual(2, graph.number_of_nodes())
         self.assertEqual(2, graph.number_of_edges())
@@ -250,8 +200,9 @@ def test_add_mirna(self):
         self.assertEqual(1, graph.number_of_nodes())
         self.assertEqual(0, graph.number_of_edges())
 
-        add_node_central_dogma(graph, mir489_gene_tuple, connection=self.manager)
-
+        result = self.manager.add_central_dogma(graph, mir489_gene_tuple)
+        self.assertIsNotNone(result)
+        self.assertIsInstance(result, mirna)
         self.assertEqual(2, graph.number_of_nodes())
         self.assertEqual(1, graph.number_of_edges())
 
@@ -265,12 +216,14 @@ def test_add_mirna(self):
     def test_add_rna(self):
         graph = BELGraph()
         mir503hg_gene = gene(namespace='HGNC', name='MIR503HG', identifier='28258')
-        graph.add_node_from_data(mir503hg_gene)
+        mir503hg_gene_tuple = graph.add_node_from_data(mir503hg_gene)
 
         self.assertEqual(1, graph.number_of_nodes())
         self.assertEqual(0, graph.number_of_edges())
 
-        add_node_central_dogma(graph, mir503hg_gene.as_tuple(), connection=self.manager)
+        result = self.manager.add_central_dogma(graph, mir503hg_gene_tuple)
+        self.assertIsNotNone(result)
+        self.assertIsInstance(result, mirna)
 
         self.assertEqual(2, graph.number_of_nodes())
         self.assertEqual(1, graph.number_of_edges())
@@ -290,7 +243,9 @@ def test_add_protein(self):
         self.assertEqual(1, graph.number_of_nodes())
         self.assertEqual(0, graph.number_of_edges())
 
-        add_node_central_dogma(graph, cd33_gene_tuple, connection=self.manager)
+        result = self.manager.add_central_dogma(graph, cd33_gene_tuple)
+        self.assertIsNotNone(result)
+        self.assertIsInstance(result, protein)
 
         self.assertEqual(3, graph.number_of_nodes())
         self.assertEqual(2, graph.number_of_edges())
@@ -329,6 +284,7 @@ def test_enrich_genes_with_families(self):
             self.assertIn(g.as_tuple(), graph.edge[cd33_gene_tuple])
 
     def test_enrich_family_with_genes(self):
+        """Test enrichment of genes members on gene families."""
         graph = BELGraph()
         f = gene(name="CD molecules", namespace=GENE_FAMILY_KEYWORD)
         graph.add_node_from_data(f)
@@ -345,7 +301,3 @@ def test_enrich_family_with_genes(self):
             g = gene(namespace='HGNC', name=x)
             self.assertTrue(graph.has_node_with_data(g))
             self.assertIn(f.as_tuple(), graph.edge[g.as_tuple()])
-
-    def test_enrich_families_with_genes(self):
-        """For gene families, adds their member genes"""
-        raise NotImplementedError