def _wr_3fmt_wrtbl(tsv_nts, goea_results, wr_params, log):
"""Using functions in the wr_tbl pkg, demonstrate printing a subset of namedtuple fields."""
from goatools.go_enrichment import get_goea_nts_prt
goea_nts = get_goea_nts_prt(goea_results)
# List of all fields, printable or not, in namedtuple (nt):
log.write("\nnamedtuple FIELDS: {F}\n".format(F=" ".join(goea_nts[0]._fields)))
# Print to Excel Spreadsheet
title = "Print subset of namedtuple fields"
wr_xlsx("nbt3102_subset_nt.xlsx", goea_nts, title=title, **wr_params)
# Print to tab-separated file
wr_tsv(tsv_nts, goea_nts, **wr_params)
def _wr_3fmt_wrtbl(goea_results, wr_params, log):
"""Using functions in the wr_tbl pkg, demonstrate printing a subset of namedtuple fields."""
from goatools.go_enrichment import get_goea_nts_prt
goea_nts = get_goea_nts_prt(goea_results)
# List of all fields, printable or not, in namedtuple (nt):
log.write("\nnamedtuple FIELDS: {F}\n".format(F=" ".join(goea_nts[0]._fields)))
# Print to Excel Spreadsheet
title = "Print subset of namedtuple fields"
wr_xlsx("nbt3102_subset_nt.xlsx", goea_nts, title=title, **wr_params)
# Print to tab-separated file
wr_tsv("nbt3102_subset_nt.tsv", goea_nts, **wr_params)
def test_combine_nt_lists():
"""Test combining lists whose elements are namedtuples or class objects."""
ntobj = cx.namedtuple("Nt", "idx")
goea_results = get_goea_results()
# Zip a list of namedtuples and another list of namedtuples
goea_nts = get_goea_nts_prt(goea_results)
lst2_nts = [ntobj._make([i]) for i in range(len(goea_nts))]
# Combine lists into a single list whose elements are a namedtuple
flds = lst2_nts[0]._fields + goea_nts[0]._fields
lst_all = combine_nt_lists([lst2_nts, goea_nts], flds)
assert lst_all[0]._fields == lst2_nts[0]._fields + goea_nts[0]._fields
# Combine list contains a subset of namedtuple fields
hdrs = ['idx', 'NS', 'level', 'depth', 'GO',
'study_count', 'study_n', 'pop_count', 'pop_n', 'p_fdr_bh', 'name']
lst_sub = combine_nt_lists([lst2_nts, goea_nts], hdrs)
assert list(lst_sub[0]._fields) == hdrs, "{F} {H}".format(F=lst_sub[0]._fields, H=hdrs)
def run_one(self, genes, tabulate=True):
if self.gene_converter is not None:
genes = self._convert(genes)
res = self.goe_obj.run_study(genes)
if self.gene_converter is not None:
# convert the study_items entry back
for t in res:
# if statement not required, but may be faster here?
if len(t.study_items) > 0:
t.study_items = self.gene_converter.index[
self.gene_converter.isin(t.study_items)].tolist()
else:
t.study_items = []
if tabulate:
prt_flds = self.goe_obj.get_prtflds_default(res)
res = pd.DataFrame(go_enrichment.get_goea_nts_prt(res, prt_flds))
res.set_index('GO', inplace=True)
return res
def test_combine_nt_lists():
"""Test combining lists whose elements are namedtuples or class objects."""
ntobj = cx.namedtuple("Nt", "idx")
goea_results = get_goea_results()
# Zip a list of namedtuples and another list of namedtuples
goea_nts = get_goea_nts_prt(goea_results)
lst2_nts = [ntobj._make([i]) for i in range(len(goea_nts))]
# Combine lists into a single list whose elements are a namedtuple
flds = lst2_nts[0]._fields + goea_nts[0]._fields
lst_all = combine_nt_lists([lst2_nts, goea_nts], flds)
assert lst_all[0]._fields == lst2_nts[0]._fields + goea_nts[0]._fields
# Combine list contains a subset of namedtuple fields
hdrs = [
'idx', 'NS', 'level', 'depth', 'GO', 'study_count', 'study_n',
'pop_count', 'pop_n', 'p_fdr_bh', 'name'
]
lst_sub = combine_nt_lists([lst2_nts, goea_nts], hdrs)
assert list(lst_sub[0]._fields) == hdrs, "{F} {H}".format(
F=lst_sub[0]._fields, H=hdrs)
def test_wrpy():
"""Test writing GOATOOLS GOEA results to a Python module as a list of nts."""
# 1. Run GOATOOLS Gene Ontology Enrichment Analysis
nature_data = get_goea_results()
# 2. Convert GOATOOLS GOEA results into a list of namedtuples
goea_results = nature_data['goea_results']
nts_goea = get_goea_nts_prt(goea_results)
# 3. Save GOATOOLS GOEA into a Python module
# 3a. Python module name
module_name = "nbt3102_goea"
fout_py = get_fout_py(module_name)
# 3b. Save GOATOOLS GOEA into a Python module
wr_py_nts(fout_py, nts_goea, varname="nts")
# 4. Check results
nts_py = importlib.import_module(module_name).nts
assert len(nts_goea) == len(nts_py)
# Alternatively, save module through goea object
module_name = "nbt3102_goea_alt"
fout_py = get_fout_py(module_name)
nature_data['goeaobj'].wr_py_goea_results(fout_py, goea_results)
nts_py = importlib.import_module(module_name).goea_results
assert len(nts_goea) == len(nts_py)
def test_wrpy():
"""Test writing GOATOOLS GOEA results to a Python module as a list of nts."""
# 1. Run GOATOOLS Gene Ontology Enrichment Analysis
nature_data = get_goea_results()
# 2. Convert GOATOOLS GOEA results into a list of namedtuples
goea_results = nature_data['goea_results']
nts_goea = get_goea_nts_prt(goea_results)
# 3. Save GOATOOLS GOEA into a Python module
# 3a. Python module name
module_name = "nbt3102_goea"
fout_py = get_fout_py(module_name)
# 3b. Save GOATOOLS GOEA into a Python module
wr_py_nts(fout_py, nts_goea, varname="nts")
# 4. Check results
nts_py = importlib.import_module(module_name).nts
assert len(nts_goea) == len(nts_py)
# Alternatively, save module through goea object
module_name = "nbt3102_goea_alt"
fout_py = get_fout_py(module_name)
nature_data['goeaobj'].wr_py_goea_results(fout_py, goea_results)
nts_py = importlib.import_module(module_name).goea_results
assert len(nts_goea) == len(nts_py)