def test_simple():
d = {}
eqs = [(1, 2), (2, 3), (4, 5)]
glom(d, eqs)
assert len(d) == 5
assert d[1] == d[2] == d[3] == {1, 2, 3}
assert d[4] == d[5] == {4, 5}
def test_sets():
d = {}
eqs = [{1, 2}, set([2, 3]), set([4, 5]), set([6, 7])]
oeqs = [{5, 7}]
glom(d, eqs)
glom(d, oeqs)
assert d[1] == d[2] == d[3] == {1, 2, 3}
assert d[4] == d[5] == d[6] == d[7] == {4, 5, 6, 7}
def test_two_calls():
d = {}
eqs = [(1, 2), (2, 3), (4, 5), (6, 7)]
oeqs = [(5, 7)]
glom(d, eqs)
glom(d, oeqs)
assert d[1] == d[2] == d[3] == {1, 2, 3}
assert d[4] == d[5] == d[6] == d[7] == {4, 5, 6, 7}
def test_bigger_sets():
d = {}
eqs = [{1, 2, 3}, {4, 5, 6}]
glom(d, eqs)
assert d[1] == d[2] == d[3] == {1, 2, 3}
assert d[4] == d[5] == d[6] == {4, 5, 6}
eqs = [{3, 4, 6, 7}]
glom(d, eqs)
assert d[1] == d[2] == d[3] == d[4] == d[5] == d[6] == d[7] == {
1, 2, 3, 4, 5, 6, 7
}
def uni_glom(unichem_data, prefix1, prefix2, chemdict):
print(f'{prefix1}/{prefix2}')
n = unichem_data.split('\n')[1:]
if len(n[-1]) == 0:
n = n[:-1]
pairs = [ni.split('\t') for ni in n]
for p in pairs:
if p[0].startswith("'") or p[1].startswith("'"):
print('UNI_GLOM {prefix1} {prefix2} {p}')
curiepairs = [(f'{prefix1}:{p[0]}', f'{prefix2}:{p[1]}') for p in pairs]
glom(chemdict, curiepairs)
def test_load_diseases_and_phenotypes(rosetta):
mondo_sets = build_sets(
rosetta.core.mondo,
['MONDO:0004979', 'MONDO:0004784', 'MONDO:0004765'])
#hpo_sets = build_sets(rosetta.core.hpo,['HP:0002099'])
dicts = {}
glom(dicts, mondo_sets)
print("*", dicts['MONDO:0004979'])
print("*", dicts['MONDO:0004784'])
print("*", dicts['MONDO:0004765'])
assert dicts['MONDO:0004979'] != dicts['MONDO:0004784'] != dicts[
'MONDO:0004765']
def load_chemicals(rosetta, refresh=False):
#Build if need be
if refresh:
refresh_mesh_pubchem(rosetta)
#Get all the simple stuff
concord = load_unichem()
#DO MESH/UNII
mesh_unii_file = os.path.join(os.path.dirname(__file__),
'mesh_to_unii.txt')
mesh_unii_pairs = load_pairs(mesh_unii_file, 'UNII')
glom(concord, mesh_unii_pairs)
#DO MESH/PUBCHEM
mesh_pc_file = os.path.join(os.path.dirname(__file__),
'mesh_to_pubchem.txt')
mesh_pc_pairs = load_pairs(mesh_pc_file, 'PUBCHEM')
glom(concord, mesh_pc_pairs)
#Dump
with open('chemconc.txt', 'w') as outf:
for key in concord:
outf.write(f'{key}\t{concord[key]}\n')
dump_cache(concord, rosetta)
def load_diseases_and_phenotypes(rosetta):
mondo_sets = build_sets(rosetta.core.mondo)
hpo_sets = build_sets(rosetta.core.hpo)
meddra_umls = read_meddra()
dicts = {}
glom(dicts, mondo_sets)
glom(dicts, hpo_sets)
glom(dicts, meddra_umls)
dump_cache(dicts, rosetta)
def load_diseases_and_phenotypes(rosetta):
print('disease/phenotype')
print('get and write mondo sets')
mondo_sets = build_exact_sets(rosetta.core.mondo,rosetta.core.uberongraph)
write_sets(mondo_sets,'mondo_sets.txt')
print('get and write hp sets')
hpo_sets = build_sets(rosetta.core.hpo, ignore_list = ['ICD','NCIT'])
hpo_sets = filter_out_non_unique_ids(hpo_sets)
write_sets(hpo_sets,'hpo_sets.txt')
print('get and write umls sets')
meddra_umls = read_meddra()
write_sets(hpo_sets,'meddra_umls_sets.txt')
dicts = {}
print('put it all together')
glom(dicts,mondo_sets)
write_dicts(dicts,'mondo_dicts.txt')
glom(dicts,hpo_sets)
write_dicts(dicts,'mondo_hpo_dicts.txt')
glom(dicts,meddra_umls)
write_dicts(dicts,'mondo_hpo_meddra_dicts.txt')
print('dump it')
with open('disease.txt','w') as outf:
dump_cache(dicts,rosetta,outf)
def load_chemicals(rosetta, refresh=True):
# Build if need be
if refresh:
refresh_mesh_pubchem(rosetta)
#Get all the simple stuff
# 1. Handle all the stuff that has an InchiKey using unichem
# 2. Mesh is all "no structure". We try to use a variety of sources to hook mesh id's to anything else
print('UNICHEM')
concord = load_unichem()
# 2. Mesh is all "no structure". We try to use a variety of sources to hook mesh id's to anything else
#DO MESH/UNII
print('MESH/UNII')
mesh_unii_file = os.path.join(os.path.dirname(__file__),
'mesh_to_unii.txt')
mesh_unii_pairs = load_pairs(mesh_unii_file, 'UNII')
glom(concord, mesh_unii_pairs)
# DO MESH/PUBCHEM
print('MESH/PUBCHEM')
mesh_pc_file = os.path.join(os.path.dirname(__file__),
'mesh_to_pubchem.txt')
mesh_pc_pairs = load_pairs(mesh_pc_file, 'PUBCHEM')
glom(concord, mesh_pc_pairs)
# DO MESH/CHEBI, but don't combine any chebi's into a set with it
print('MESH/CHEBI')
mesh_chebi = pull_mesh_chebi()
#Merging CHEBIS can be ok because of primary/secondary chebis. Really we
# don't want to merge INCHIs
#glom(concord, mesh_chebi,['INCHI'])
glom(concord, mesh_chebi)
# 3. Pull from chebi the sdf and db files, use them to link to things (KEGG) in the no inchi/no smiles cases
pubchem_chebi_pairs, kegg_chebi_pairs = pull_chebi()
glom(concord, pubchem_chebi_pairs)
glom(concord, kegg_chebi_pairs)
# 4. Go to KEGG, and get sequences for peptides.
sequence_concord = rosetta.core.kegg.pull_sequences()
# 5. Pull UniProt (swissprot) XML.
# Calculate sequences for the sub-sequences (Uniprot_PRO)
sequence_to_uniprot = pull_uniprot(refresh)
# 6. Use the sequences to merge UniProt with KEGG
for s, v in sequence_to_uniprot.items():
sequence_concord[s].update(v)
# 7. Read IUPHAR, discard things with INCHI, use things with sequence to match UniProt_PRO/KEGG
# Use the hand-curated version of IUPHAR to match the un-sequenced stuff left over
sequence_to_iuphar, iuphar_glom = pull_iuphar()
for s, v in sequence_to_iuphar.items():
sequence_concord[s].update(v)
glom(concord, iuphar_glom)
# 8. Use wikidata to get links between CHEBI and UniProt_PRO
unichebi = pull_uniprot_chebi()
glom(concord, unichebi)
# 9. glom across sequence and chemical stuff
new_groups = sequence_concord.values()
glom(concord, new_groups, unique_prefixes=['GTOPDB', 'INCHI'])
# 10. Drop PRO only sequences.
to_remove = []
for eq_id_set in concord:
if len(eq_id_set) > 1:
continue
print(eq_id_set)
item = iter(eq_id_set).next()
if '#PRO_' in item:
to_remove.add(eq_id_set)
for eids in to_remove:
concord.remove(eids)
#Add labels to CHEBIs, CHEMBLs, MESHes
print('LABEL')
label_chebis(concord)
label_chembls(concord, refresh=refresh)
label_meshes(concord)
print('dumping')
#Dump
with open('chemconc.txt', 'w') as outf:
dump_cache(concord, rosetta, outf)
#dump_cache(concord,rosetta)
print('done')