Upgrade to PyBEL 14 / Bio2BEL 3 (#10)

.bumpversion.cfg

@@ -1,5 +1,5 @@
 [bumpversion]
-current_version = 0.2.5-dev
+current_version = 0.3.0-dev
 commit = True
 tag = False
 parse = (?P<major>\d+)\.(?P<minor>\d+)\.(?P<patch>\d+)(?:-(?P<release>[0-9A-Za-z-]+(?:\.[0-9A-Za-z-]+)*))?(?:\+(?P<build>[0-9A-Za-z-]+(?:\.[0-9A-Za-z-]+)*))?

.readthedocs.yml

@@ -1,8 +1,8 @@
 # See: https://docs.readthedocs.io/en/latest/yaml-config.html
 build:
-    image: latest
+  image: latest
 python:
   version: 3.6
   pip_install: true
   extra_requirements:
-  - docs
+    - docs

.travis.yml

@@ -2,7 +2,7 @@ sudo: false
 cache: pip
 language: python
 python:
-- 3.6
+  - '3.7'
 stages:
   - lint
   - docs
@@ -15,7 +15,6 @@ jobs:
     - env: TOXENV=flake8
     - env: TOXENV=pyroma
     - env: TOXENV=xenon
-    - env: TOXENV=pyroma
     # docs stage
     - stage: docs
       env: TOXENV=doc8

docs/source/conf.py

@@ -54,7 +54,7 @@
 copyright = '2017-2018, Charles Tapley Hoyt'
 author = 'Charles Tapley Hoyt'
 
-release = '0.2.5-dev'
+release = '0.3.0-dev'
 
 parsed_version = re.match(
     '(?P<major>\d+)\.(?P<minor>\d+)\.(?P<patch>\d+)(?:-(?P<release>[0-9A-Za-z-]+(?:\.[0-9A-Za-z-]+)*))?(?:\+(?P<build>[0-9A-Za-z-]+(?:\.[0-9A-Za-z-]+)*))?',

setup.py

@@ -24,8 +24,8 @@
     'Topic :: Scientific/Engineering :: Bio-Informatics',
 ]
 INSTALL_REQUIRES = [
-    'pybel>=0.13.1,<0.14.0',
-    'bio2bel>=0.2.0,<0.3.0',
+    'pybel>=0.14.0,<0.15.0',
+    'bio2bel>=0.3.0,<0.4.0',
     'tqdm',
     'sqlalchemy',
     'click',

src/bio2bel_hgnc/__init__.py

@@ -6,14 +6,14 @@
 from .manager import Manager  # noqa: F401
 from .utils import get_version  # noqa: F401
 
-__version__ = '0.2.5-dev'
+__version__ = '0.3.0-dev'
 
 __title__ = 'bio2bel_hgnc'
 __description__ = "A package for converting HGNC to BEL"
 __url__ = 'https://github.com/bio2bel/hgnc'
 
 __author__ = 'Charles Tapley Hoyt'
-__email__ = 'charles.hoyt@scai.fraunhofer.de'
+__email__ = 'cthoyt@gmail.com'
 
 __license__ = 'MIT License'
-__copyright__ = 'Copyright (c) 2017-2018 Charles Tapley Hoyt'
+__copyright__ = 'Copyright (c) 2017-2019 Charles Tapley Hoyt'

src/bio2bel_hgnc/cli.py

@@ -2,12 +2,8 @@
 
 """Command line interface for Bio2BEL HGNC."""
 
-import logging
-
 from .manager import Manager
 
-log = logging.getLogger(__name__)
-
 main = Manager.get_cli()
 
 if __name__ == '__main__':

src/bio2bel_hgnc/constants.py

@@ -4,7 +4,7 @@
 
 from bio2bel import get_data_dir
 
-VERSION = '0.2.5-dev'
+VERSION = '0.3.0-dev'
 
 MODULE_NAME = 'hgnc'
 #: The default directory where PyBEL files, including logs and the  default cache, are stored. Created if not exists.

src/bio2bel_hgnc/gfam_manager.py

@@ -13,9 +13,7 @@
 from pybel import BELGraph
 from pybel.manager.models import Namespace, NamespaceEntry
 from .model_utils import family_to_bel, gene_to_bel
-from .models import (
-    Base, GeneFamily, gene_gene_family,
-)
+from .models import Base, GeneFamily, gene_gene_family
 from .wrapper import BaseManager
 
 __all__ = [

src/bio2bel_hgnc/manager.py

@@ -16,7 +16,7 @@
 from bio2bel.manager.flask_manager import FlaskMixin
 from bio2bel.manager.namespace_manager import BELNamespaceManagerMixin
 from pybel import BELGraph
-from pybel.constants import FUNCTION, GENE, IDENTIFIER, MIRNA, NAME, NAMESPACE, PROTEIN, RNA, VARIANTS
+from pybel.constants import FUNCTION, MIRNA, NAME, NAMESPACE, PROTEIN, RNA, VARIANTS
 from pybel.dsl import BaseEntity, CentralDogma, FUNC_TO_DSL, rna as rna_dsl
 from pybel.manager.models import Namespace, NamespaceEntry
 from .constants import ENCODINGS, ENTREZ, MODULE_NAME
@@ -28,14 +28,13 @@
 )
 from .wrapper import BaseManager
 
-log = logging.getLogger(__name__)
-
 __all__ = [
     'Manager',
 ]
 
-UNIPROT_RE = r'^([A-N,R-Z][0-9]([A-Z][A-Z, 0-9][A-Z, 0-9][0-9]){1,2})|([O,P,Q][0-9][A-Z, 0-9][A-Z, 0-9][A-Z, 0-9][0-9])(\.\d+)?$'
+logger = logging.getLogger(__name__)
 
+UNIPROT_RE = r'^([A-N,R-Z][0-9]([A-Z][A-Z, 0-9][A-Z, 0-9][0-9]){1,2})|([O,P,Q][0-9][A-Z, 0-9][A-Z, 0-9][A-Z, 0-9][0-9])(\.\d+)?$'
 GENE_FAMILY_NAMESPACES = {'gfam', 'hgnc.family', 'hgnc.genefamily'}
 
 
@@ -181,7 +180,7 @@ def get_gene_by_mgi_id(self, mgi_id: str) -> Optional[HumanGene]:
         human_genes = mouse_gene.hgncs
 
         if len(human_genes) > 1:
-            log.warning('multiple human genes mapped to mgi_id:%s: %s', mgi_id, human_genes)
+            logger.warning('multiple human genes mapped to mgi_id:%s: %s', mgi_id, human_genes)
             return
 
         return human_genes[0]
@@ -200,7 +199,7 @@ def get_gene_by_rgd_id(self, rgd_id: str) -> Optional[HumanGene]:
         human_genes = rat_gene.hgncs
 
         if len(human_genes) > 1:
-            log.warning('multiple human genes mapped to rgd_id:%s: %s', rgd_id, human_genes)
+            logger.warning('multiple human genes mapped to rgd_id:%s: %s', rgd_id, human_genes)
             return
 
         return human_genes[0]
@@ -230,12 +229,15 @@ def get_node(self, node: BaseEntity) -> Optional[HumanGene]:
         :param node: The node to look for
         :raises: KeyError
         """
-        if NAMESPACE not in node:
+        if not isinstance(node, CentralDogma):
+            return
+
+        namespace = node.namespace
+        if namespace is None:
             return
 
-        namespace = node[NAMESPACE]
-        identifier = node.get(IDENTIFIER)
-        name = node.get(NAME)
+        identifier = node.identifier
+        name = node.name
 
         if namespace.lower() in {'hgnc'}:
             return self._get_node_handle_hgnc(identifier, name)
@@ -325,26 +327,22 @@ def normalize_genes(self, graph: BELGraph, use_tqdm: bool = False) -> None:
 
     def enrich_genes_with_equivalences(self, graph: BELGraph) -> None:
         """Enrich genes with their corresponding UniProt."""
-        self.add_namespace_to_graph(graph)
-
-        if 'uniprot' not in graph.namespace_url:
-            graph.namespace_pattern['uniprot'] = UNIPROT_RE
-
         for node, human_gene in list(self.iter_genes(graph)):
             func = node.function
 
             if human_gene.entrez:
-                graph.add_equivalence(node, FUNC_TO_DSL[func](
+                entrez_node = FUNC_TO_DSL[func](
                     namespace=ENTREZ,
                     name=human_gene.symbol,
                     identifier=str(human_gene.entrez)
-                ))
+                )
+                graph.add_equivalence(node, entrez_node)
 
-            if func in {PROTEIN, RNA, GENE}:
+            if func == PROTEIN:
                 for uniprot in human_gene.uniprots:
                     graph.add_equivalence(node, uniprot_to_bel(uniprot))
 
-            if func in {RNA, GENE}:
+            if func == RNA:
                 if human_gene.mirbase:
                     mirbase_rna_node = rna_dsl(
                         namespace='mirbase',
@@ -397,21 +395,25 @@ def enrich_families_with_genes(self, graph: BELGraph):
         self.add_namespace_to_graph(graph)
 
         for gene_family_node in list(graph):
-            if gene_family_node[FUNCTION] != GENE:
+            if not isinstance(gene_family_node, pybel.dsl.Gene):
                 continue
 
-            if gene_family_node.get(NAMESPACE).lower() not in GENE_FAMILY_NAMESPACES:
+            namespace = gene_family_node.namespace
+            if namespace is None or namespace.lower() not in GENE_FAMILY_NAMESPACES:
                 continue
 
-            if IDENTIFIER in gene_family_node:
-                gene_family_model = self.get_family_by_id(gene_family_node[IDENTIFIER])
-            elif NAME in gene_family_node:
-                gene_family_model = self.get_family_by_name(gene_family_node[NAME])
+            identifier = gene_family_node.identifier
+            name = gene_family_node.name
+
+            if identifier:
+                gene_family_model = self.get_family_by_id(identifier)
+            elif name:
+                gene_family_model = self.get_family_by_name(name)
             else:
                 raise ValueError
 
             if gene_family_model is None:
-                log.info('family not found: %s', gene_family_node)
+                logger.info('family not found: %s', gene_family_node)
                 continue
 
             for human_gene in gene_family_model.hgncs:
@@ -523,12 +525,12 @@ def to_bel(self) -> BELGraph:
         )
 
         hgnc_namespace = self.upload_bel_namespace()
-        log.info('using default namespace: %s at %s', hgnc_namespace, hgnc_namespace.url)
+        logger.info('using default namespace: %s at %s', hgnc_namespace, hgnc_namespace.url)
         graph.namespace_url[hgnc_namespace.keyword] = hgnc_namespace.url
 
         gfam_manager = GfamManager(connection=self.connection)
         gfam_namespace = gfam_manager.upload_bel_namespace()
-        log.info('using default namespace: %s at %s', gfam_namespace, gfam_namespace.url)
+        logger.info('using default namespace: %s at %s', gfam_namespace, gfam_namespace.url)
         graph.namespace_url[gfam_namespace.keyword] = gfam_namespace.url
 
         for human_gene in tqdm(self.list_human_genes(), total=self.count_human_genes(),
@@ -613,9 +615,9 @@ def _cli_add_populate(main: click.Group) -> click.Group:  # noqa: D202
         def populate(manager, reset, skip_hcop):
             """Populate the database."""
             if reset:
-                log.info('Deleting the previous instance of the database')
+                logger.info('Deleting the previous instance of the database')
                 manager.drop_all()
-                log.info('Creating new models')
+                logger.info('Creating new models')
                 manager.create_all()
 
             manager.populate(use_hcop=(not skip_hcop))

src/bio2bel_hgnc/model_utils.py

@@ -17,8 +17,11 @@
 ]
 
 
-def gene_to_bel(human_gene: HumanGene, func: Optional[str] = None,
-                variants: Optional[List[Variant]] = None) -> CentralDogma:
+def gene_to_bel(
+    human_gene: HumanGene,
+    func: Optional[str] = None,
+    variants: Optional[List[Variant]] = None,
+) -> CentralDogma:
     """Convert a Gene to a PyBEL gene."""
     dsl = FUNC_TO_DSL[func] if func else gene_dsl
 
@@ -34,8 +37,11 @@ def gene_to_bel(human_gene: HumanGene, func: Optional[str] = None,
     return rv
 
 
-def family_to_bel(family: GeneFamily, func: Optional[str] = None,
-                  variants: Optional[List[Variant]] = None) -> CentralDogma:
+def family_to_bel(
+    family: GeneFamily,
+    func: Optional[str] = None,
+    variants: Optional[List[Variant]] = None,
+) -> CentralDogma:
     """Convert a Gene Family model to a PyBEL gene."""
     dsl = FUNC_TO_DSL[func] if func else gene_dsl
 

tests/test_enrich_metadata.py

@@ -5,6 +5,7 @@
 import logging
 import unittest
 
+import pybel.dsl
 from bio2bel_hgnc import Manager
 from bio2bel_hgnc.constants import ENTREZ, HGNC, HGNC_GENE_FAMILY
 from bio2bel_hgnc.models import HumanGene
@@ -13,7 +14,7 @@
 from pybel.dsl import gene, mirna, protein, rna
 from tests.cases import TemporaryCacheMixin
 
-log = logging.getLogger(__name__)
+logger = logging.getLogger(__name__)
 
 protein_hgnc_cd33 = protein(name='CD33', namespace=HGNC)
 gene_hgnc_cd33 = protein_hgnc_cd33.get_rna().get_gene()
@@ -74,35 +75,37 @@ def test_get_rgd_node(self):
         cd33_model = self.manager.get_node(cd33_rgd_name)
         self.help_check_cd33_model(cd33_model)
 
-    def test_get_rgd_id_node(self):
-        """Test getting a node by RGD identifier."""
-        cd33_model = self.manager.get_node(cd33_rgd_id)
-        self.help_check_cd33_model(cd33_model)
-
     @unittest.skip('HGNC does not have MGI symbol information')
     def test_get_mgi_node(self):
         """Test getting a node by MGI name."""
         cd33_model = self.manager.get_node(cd33_mgi_name)
         self.help_check_cd33_model(cd33_model)
 
-    def test_get_mgi_id_node(self):
-        """Test getting a node by MGI identifier."""
-        cd33_model = self.manager.get_node(cd33_mgi_identifier)
-        self.help_check_cd33_model(cd33_model)
-
     def test_get_entrez_node(self):
         """Test getting a node by Entrez Gene identifier."""
         cd33_model = self.manager.get_node(cd33_entrez)
         self.help_check_cd33_model(cd33_model)
 
     def test_add_equivalency(self):
         """Test that CD33 identified by HGNC and entrez can be equivalenced."""
+        gene_model = self.manager.get_gene_by_hgnc_symbol('CD33')
+        self.assertIsNotNone(gene_model)
+        self.assertEqual('CD33', gene_model.symbol)
+        self.assertEqual('1659', str(gene_model.identifier))
+        self.assertEqual('945', gene_model.entrez)
+
         graph = BELGraph()
         graph.add_node_from_data(protein_hgnc_cd33)
 
         self.assertEqual(1, graph.number_of_nodes(), msg='wrong initial number of nodes')
         self.assertEqual(0, graph.number_of_edges(), msg='wrong initial number of edges')
 
+        genes = list(self.manager.iter_genes(graph))
+        self.assertEqual(1, len(genes))
+        e, m = genes[0]
+        self.assertIsInstance(e, pybel.dsl.Protein)
+        self.assertIsInstance(m, HumanGene)
+
         self.manager.enrich_genes_with_equivalences(graph)
 
         # self.assertEqual(2, graph.number_of_nodes(), msg='nodes: {}'.format(list(graph)))

tox.ini

@@ -42,12 +42,6 @@ deps = check-manifest
 commands = check-manifest
 skip_install = true
 
-[testenv:vulture]
-deps = vulture
-skip_install = true
-commands = vulture src/bio2bel_hgnc/
-description = Run the vulture tool to look for dead code.
-
 [testenv:xenon]
 deps = xenon
 skip_install = true
@@ -66,7 +60,7 @@ deps =
     pyroma
 skip_install = true
 commands = pyroma --min=10 .
-description = Run the pyroma tool to check the project's package friendliness.
+description = Run the pyroma tool to check the project package friendliness.
 
 [testenv:docs]
 changedir = docs