def __init__(self, taxid, fin_gene2go, fin_gobasic):
_fin = os.path.join(REPO, fin_gene2go)
dnld_ncbi_gene_file(_fin, loading_bar=None)
self.gene2go = read_ncbi_gene2go(_fin, [taxid])
_fin_obo = os.path.join(REPO, fin_gobasic)
self.godag = get_godag(_fin_obo, loading_bar=None)
def __init__(self, taxid, fin_gene2go, fin_gobasic):
_fin = os.path.join(REPO, fin_gene2go)
dnld_ncbi_gene_file(_fin, loading_bar=None)
self.gene2go = read_ncbi_gene2go(_fin, [taxid])
_fin_obo = os.path.join(REPO, fin_gobasic)
self.godag = get_godag(_fin_obo, loading_bar=None)
def _dnld_anno(file_anno):
"""Download the annotation file, if needed."""
if os.path.exists(file_anno):
assert os.path.getsize(file_anno) > 1000000, "BAD ANNO({F})".format(F=file_anno)
return
dnld_ncbi_gene_file(file_anno, loading_bar=None)
assert os.path.isfile(file_anno), "MISSING ANNO({F})".format(F=file_anno)
assert os.path.getsize(file_anno) > 1000000, "BAD ANNO({F})".format(F=file_anno)
def _get_id2gos(file_assc, taxids, log):
"""Return associations."""
taxid2asscs = defaultdict(lambda: defaultdict(lambda: defaultdict(set)))
fin = os.path.join(REPO, file_assc)
dnld_ncbi_gene_file(fin, loading_bar=None)
id2gos = read_ncbi_gene2go(fin, taxids, taxid2asscs=taxid2asscs)
log.write(" {N} items found in gene2go from NCBI's ftp server\n".format(N=len(id2gos)))
return taxid2asscs
def _dnld_anno(file_anno):
"""Download the annotation file, if needed."""
if os.path.exists(file_anno):
assert os.path.getsize(file_anno) > 1000000, "BAD ANNO({F})".format(F=file_anno)
return
dnld_ncbi_gene_file(file_anno, loading_bar=None)
assert os.path.isfile(file_anno), "MISSING ANNO({F})".format(F=file_anno)
assert os.path.getsize(file_anno) > 1000000, "BAD ANNO({F})".format(F=file_anno)
def _get_evidencecodes(fin_gene2go):
"""Get all evidence codes and qualifiers."""
evs = set()
fin_gene2go = os.path.join(REPO, 'gene2go')
dnld_ncbi_gene_file(fin_gene2go, force_dnld=False, loading_bar=False)
with open(fin_gene2go) as ifstrm:
for line in ifstrm:
if line[0] != '#': # Line contains data. Not a comment
line = line.rstrip() # chomp
flds = line.split('\t')
if len(flds) >= 5:
# taxid_curr, geneid, go_id, evidence, qualifier = flds[:5]
evidence = flds[3]
assert len(evidence) >= 2, flds
evs.add(evidence)
print('{N} evidence codes in {FIN}'.format(N=len(evs), FIN=fin_gene2go))
return evs
def _get_evidencecodes(fin_gene2go):
"""Get all evidence codes and qualifiers."""
evs = set()
fin_gene2go = os.path.join(REPO, 'gene2go')
dnld_ncbi_gene_file(fin_gene2go, force_dnld=False, loading_bar=False)
with open(fin_gene2go) as ifstrm:
for line in ifstrm:
if line[0] != '#': # Line contains data. Not a comment
line = line.rstrip() # chomp
flds = line.split('\t')
if len(flds) >= 5:
# taxid_curr, geneid, go_id, evidence, qualifier = flds[:5]
evidence = flds[3]
assert len(evidence) >= 2, flds
evs.add(evidence)
print('{N} evidence codes in {FIN}'.format(N=len(evs), FIN=fin_gene2go))
return evs
def test_i96():
"""Test to re-produce issue#96: Passes currently."""
# Trying to duplicate: ValueError("All values in table must be nonnegative.
# Get genes
print('CWD', os.getcwd())
study_ids = _get_geneids()
population_ids = GENEID2NT.keys()
# Get databases
print(os.getcwd())
fin = os.path.join(REPO, 'gene2go')
dnld_ncbi_gene_file(fin, loading_bar=None)
gene2go = read_ncbi_gene2go(fin, [9606])
fin_obo = os.path.join(REPO, "go-basic.obo")
godag = get_godag(fin_obo, loading_bar=None)
goeaobj = GOEnrichmentStudy(population_ids, gene2go, godag, methods=['fdr_bh'])
# Run GOEA Gene Ontology Enrichment Analysis
results_goeas = goeaobj.run_study(study_ids)
def test_i96():
"""Test to re-produce issue#96: Passes currently."""
# Trying to duplicate: ValueError("All values in table must be nonnegative.
# Get genes
print('CWD', os.getcwd())
study_ids = _get_geneids()
population_ids = GENEID2NT.keys()
# Get databases
print(os.getcwd())
fin = os.path.join(REPO, 'gene2go')
dnld_ncbi_gene_file(fin, loading_bar=None)
gene2go = read_ncbi_gene2go(fin, [9606])
fin_obo = os.path.join(REPO, "go-basic.obo")
godag = get_godag(fin_obo, loading_bar=None)
goeaobj = GOEnrichmentStudy(population_ids,
gene2go,
godag,
methods=['fdr_bh'])
# Run GOEA Gene Ontology Enrichment Analysis
results_goeas = goeaobj.run_study(study_ids)
