def get_bel() -> pybel.BELGraph:
"""Get the ComPath mappings as BEL."""
graph = pybel.BELGraph(
name='ComPath Mappings',
version='1.1.0',
description=
'Hierarchical and equivalence relations between entries in KEGG, Reactome, PathBank,'
' and WikiPathways.')
df = get_df()
for source_ns, source_id, source_name, relation, target_ns, target_id, target_name in df.values:
source = BiologicalProcess(
namespace=source_ns,
identifier=source_id,
name=source_name,
)
target = BiologicalProcess(
namespace=target_ns,
identifier=target_id,
name=target_name,
)
if relation == 'isPartOf':
graph.add_part_of(source, target)
elif relation == 'equivalentTo':
graph.add_equivalence(source, target)
else:
raise ValueError(f'invalid mapping with relation: {relation}')
return graph
def gobp(name: str, identifier: str) -> BiologicalProcess:
"""Make a GO biological process node."""
return BiologicalProcess(
namespace='go',
name=name,
identifier=identifier,
)
def test_bioprocess(self):
node = BiologicalProcess(namespace='GO', name='apoptosis')
self.assertEqual('bp(GO:apoptosis)', str(node))
Gene('HGNC', 'NCF1'),
Protein('HGNC', 'HBP1')
]),
Protein('HGNC', 'HBP1'),
ComplexAbundance([Protein('HGNC', 'FOS'), Protein('HGNC', 'JUN')]),
Protein('HGNC', 'FOS'),
Protein('HGNC', 'JUN'),
Rna('HGNC', 'CFTR', variants=Hgvs('r.1521_1523delcuu')),
Rna('HGNC', 'CFTR'),
Rna('HGNC', 'CFTR', variants=Hgvs('r.1653_1655delcuu')),
CompositeAbundance([
interleukin_23_complex,
il6
]),
il6,
BiologicalProcess('GO', 'cell cycle arrest'),
hydrogen_peroxide,
Protein('HGNC', 'CAT'),
Gene('HGNC', 'CAT'),
Protein('HGNC', 'HMGCR'),
BiologicalProcess('GO', 'cholesterol biosynthetic process'),
Gene('HGNC', 'APP', variants=Hgvs('c.275341G>C')),
Gene('HGNC', 'APP'),
Pathology('MESHD', 'Alzheimer Disease'),
ComplexAbundance([Protein('HGNC', 'F3'), Protein('HGNC', 'F7')]),
Protein('HGNC', 'F3'),
Protein('HGNC', 'F7'),
Protein('HGNC', 'F9'),
Protein('HGNC', 'GSK3B', variants=ProteinModification('Ph', 'Ser', 9)),
Protein('HGNC', 'GSK3B'),
Pathology('MESHD', 'Psoriasis'),
ComplexAbundance,
NamedComplexAbundance,
Pathology,
Protein,
ProteinModification,
)
from pybel.language import activity_mapping
from pybel.testing.constants import test_jgif_path
from tests.constants import TestGraphMixin
logging.getLogger('pybel.parser').setLevel(20)
calcium = Abundance('SCHEM', 'Calcium')
calcineurin_complex = NamedComplexAbundance('SCOMP', 'Calcineurin Complex')
foxo3 = Protein('HGNC', 'FOXO3')
tcell_proliferation = BiologicalProcess(
'GO', 'CD8-positive, alpha-beta T cell proliferation')
il15 = Protein('HGNC', 'IL15')
il2rg = Protein('MGI', 'Il2rg')
jgif_expected_nodes = {
calcium,
calcineurin_complex,
foxo3,
tcell_proliferation,
il15,
il2rg,
Protein('HGNC', 'CXCR6'),
Protein('HGNC', 'IL15RA'),
BiologicalProcess('GO', 'lymphocyte chemotaxis'),
Protein('HGNC', 'IL2RG'),
Protein('HGNC', 'ZAP70'),
NamedComplexAbundance('SCOMP', 'T Cell Receptor Complex'),
return {RELATION: x}
def _rela(x, y=None):
return {RELATION: x, OBJECT: activity(y)}
def _assoc(y):
return {RELATION: ASSOCIATION, 'association_type': y}
a1 = Abundance('CHEBI', '1')
p1 = Protein('HGNC', '1')
pf1 = Protein('INTERPRO', '1')
d1 = Pathology('MESH', '1')
b1 = BiologicalProcess('GO', '1')
b2 = BiologicalProcess('GO', '2')
m1 = MicroRna('MIRBASE', '1')
r1 = Rna('HGNC', '1')
r2 = Rna('HGNC', '2')
nca1 = NamedComplexAbundance('FPLX', '1')
pop1 = Population('taxonomy', '1')
p2 = Protein('HGNC', identifier='9236')
p3 = Protein('HGNC', identifier='9212')
r3 = p3.get_rna()
g3 = r3.get_gene()
c1 = ComplexAbundance([p2, g3])
c2 = ComplexAbundance([p1, p2])
c3 = ComplexAbundance([a1, p2])
# PyBEL manager
cls.pybel_manager = pybel.Manager(engine=cls.engine,
session=cls.session)
cls.pybel_manager.create_all()
@classmethod
def tearDownClass(cls):
"""Close the connection in the manager and deletes the temporary database."""
cls.session.close()
super().tearDownClass()
protein_a = Protein(namespace=HGNC, identifier='2976', name='DNMT1')
protein_b = Protein(namespace=HGNC, identifier='9173', name='POLA1')
gene_c = Gene(namespace=HGNC, identifier='8903', name='PGLS')
pathway_a = BiologicalProcess(namespace=WIKIPATHWAYS,
identifier='WP1604',
name='Codeine and Morphine Metabolism')
def get_enrichment_graph():
"""Build a simple test graph with 2 proteins, one gene, and one pathway all contained in HGNC."""
graph = BELGraph(
name='My test graph for enrichment',
version='0.0.1',
)
graph.add_increases(protein_a, protein_b, citation='1234', evidence='')
graph.add_decreases(protein_b, gene_c, citation='1234', evidence='')
graph.add_part_of(gene_c, pathway_a)
return graph
def normalize_graph_names(graph: BELGraph, database: str) -> None:
"""Normalize graph names."""
# Victim to Survivor (one to one node) mapping
one_to_one_mapping = {}
# Victim to Survivors (one to many nodes) mapping
one_to_many_mapping = defaultdict(set)
for node in graph.nodes():
# Skip ListAbundances and Reactions since they do not have a name
if isinstance(node, ListAbundance) or isinstance(
node, Reaction) or not node.name:
continue
# Normalize names: Lower case name and strip quotes or white spaces
lower_name = node.name.lower().strip('"').strip()
# Dealing with Genes/miRNAs
if isinstance(node, CentralDogma):
##################
# miRNA entities #
##################
if lower_name.startswith("mir"):
# Reactome preprocessing to flat multiple identifiers
if database == REACTOME:
reactome_cell = munge_reactome_gene(lower_name)
if isinstance(reactome_cell, list):
for lower_name in reactome_cell:
one_to_many_mapping[node].add(
MicroRna(
node.namespace,
name=lower_name.replace("mir-", "mir"),
identifier=node.identifier,
), )
if lower_name.endswith(' genes'):
lower_name = lower_name[:-len(' genes')]
elif lower_name.endswith(' gene'):
lower_name = lower_name[:-len(' gene')]
one_to_one_mapping[node] = MicroRna(
node.namespace,
name=lower_name.replace(
"mir-", "mir"), # Special case for Reactome
)
continue
# KEGG and Reactome
one_to_one_mapping[node] = MicroRna(
node.namespace,
name=node.name.replace("mir-", "mir"),
identifier=node.identifier,
)
##################
# Genes entities #
##################
else:
# Reactome preprocessing to flat multiple identifiers
if database == REACTOME:
reactome_cell = munge_reactome_gene(lower_name)
if isinstance(reactome_cell, list):
for lower_name in reactome_cell:
if lower_name in BLACK_LIST_REACTOME: # Filter entities in black list
continue
elif lower_name.startswith(
"("): # remove redundant parentheses
lower_name = lower_name.strip("(").strip(")")
one_to_many_mapping[node].add(
Protein(node.namespace,
name=lower_name,
identifier=node.identifier), )
else:
one_to_one_mapping[node] = Protein(
node.namespace,
name=lower_name,
identifier=node.identifier)
continue
# WikiPathways and KEGG do not require any processing of genes
elif database == WIKIPATHWAYS and lower_name in WIKIPATHWAYS_BIOL_PROCESS:
one_to_one_mapping[node] = BiologicalProcess(
node.namespace,
name=lower_name,
identifier=node.identifier,
)
continue
one_to_one_mapping[node] = Protein(node.namespace,
name=lower_name,
identifier=node.identifier)
#######################
# Metabolite entities #
#######################
elif isinstance(node, Abundance):
if database == 'wikipathways':
# Biological processes that are captured as abundance in
# BEL since they were characterized wrong in WikiPathways
if lower_name in WIKIPATHWAYS_BIOL_PROCESS:
one_to_one_mapping[node] = BiologicalProcess(
node.namespace,
name=lower_name,
identifier=node.identifier,
)
continue
# Abundances to BiologicalProcesses
elif (node.namespace
in {'WIKIDATA', 'WIKIPATHWAYS', 'REACTOME'}
and lower_name not in WIKIPATHWAYS_METAB):
one_to_one_mapping[node] = BiologicalProcess(
node.namespace,
name=lower_name,
identifier=node.identifier,
)
continue
# Fix naming in duplicate entity
if lower_name in WIKIPATHWAYS_NAME_NORMALIZATION:
lower_name = WIKIPATHWAYS_NAME_NORMALIZATION[lower_name]
elif database == REACTOME:
# Curated proteins that were coded as metabolites
if lower_name in REACTOME_PROT:
one_to_one_mapping[node] = Protein(
node.namespace,
name=lower_name,
identifier=node.identifier,
)
continue
# Flat multiple identifiers (this is not trivial because most of ChEBI names contain commas,
# so a clever way to fix some of the entities is to check that all identifiers contain letters)
elif "," in lower_name and all(
string.isalpha() for string in lower_name.split(",")):
for string in lower_name.split(","):
one_to_many_mapping[node].add(
Abundance(node.namespace,
name=string,
identifier=node.identifier), )
continue
one_to_one_mapping[node] = Abundance(node.namespace,
name=lower_name,
identifier=node.identifier)
#################################
# Biological Processes entities #
#################################
elif isinstance(node, BiologicalProcess):
# KEGG normalize name by removing the title prefix
if lower_name.startswith('title:'):
lower_name = lower_name[len('title:'):]
one_to_one_mapping[node] = BiologicalProcess(
node.namespace,
name=lower_name,
identifier=node.identifier,
)
relabel_nodes(graph, one_to_one_mapping)
multi_relabel(graph, one_to_many_mapping)
'AKT1',
variants=[Hgvs('c.1521_1523delCTT'),
Hgvs('p.Phe508del')]),
Gene('HGNC', 'NCF1'),
ComplexAbundance([Gene('HGNC', 'NCF1'),
Protein('HGNC', 'HBP1')]),
Protein('HGNC', 'HBP1'),
ComplexAbundance([Protein('HGNC', 'FOS'),
Protein('HGNC', 'JUN')]),
Protein('HGNC', 'FOS'),
Protein('HGNC', 'JUN'),
Rna('HGNC', 'CFTR', variants=Hgvs('r.1521_1523delcuu')),
Rna('HGNC', 'CFTR'),
Rna('HGNC', 'CFTR', variants=Hgvs('r.1653_1655delcuu')),
CompositeAbundance([interleukin_23_complex, il6]), il6,
BiologicalProcess('GO', 'cell cycle arrest'), hydrogen_peroxide,
Protein('HGNC', 'CAT'),
Gene('HGNC', 'CAT'),
Protein('HGNC', 'HMGCR'),
BiologicalProcess('GO', 'cholesterol biosynthetic process'),
Gene('HGNC', 'APP', variants=Hgvs('c.275341G>C')),
Gene('HGNC', 'APP'),
Pathology('MESHD', 'Alzheimer Disease'),
ComplexAbundance([Protein('HGNC', 'F3'),
Protein('HGNC', 'F7')]),
Protein('HGNC', 'F3'),
Protein('HGNC', 'F7'),
Protein('HGNC', 'F9'),
Protein('HGNC', 'GSK3B', variants=ProteinModification('Ph', 'Ser', 9)),
Protein('HGNC', 'GSK3B'),
Pathology('MESHD', 'Psoriasis'),