def run_study(self, study, **kws):
"""Run Gene Ontology Enrichment Study (GOEA) on study ids."""
# Key-word arguments:
methods = Methods(kws['methods']) if 'methods' in kws else self.methods
alpha = kws['alpha'] if 'alpha' in kws else self.alpha
log = kws['log'] if 'log' in kws else self.log
# Calculate uncorrected pvalues
results = self.get_pval_uncorr(study, log)
if not results:
return []
if log is not None:
log.write(" {MSG}\n".format(MSG="\n ".join(self.get_results_msg(results, study))))
# Do multipletest corrections on uncorrected pvalues and update results
self._run_multitest_corr(results, methods, alpha, study, log)
for rec in results:
# get go term for name and level
rec.set_goterm(self.obo_dag)
# 'keep_if' can be used to keep only significant GO terms. Example:
# >>> keep_if = lambda nt: nt.p_fdr_bh < 0.05 # if results are significant
# >>> goea_results = goeaobj.run_study(geneids_study, keep_if=keep_if)
if 'keep_if' in kws:
keep_if = kws['keep_if']
results = [r for r in results if keep_if(r)]
# Default sort order: First, sort by BP, MF, CC. Second, sort by pval
results.sort(key=lambda r: [r.NS, r.enrichment, r.p_uncorrected])
return results # list of GOEnrichmentRecord objects
def __init__(self,
pop,
assoc,
obo_dag,
propagate_counts=True,
alpha=.05,
methods=None,
**kws):
self.name = kws.get('name', 'GOEA')
print('\nLoad {OBJNAME} Gene Ontology Analysis ...'.format(
OBJNAME=self.name))
self.log = kws['log'] if 'log' in kws else sys.stdout
self._run_multitest = {
'local': self._run_multitest_local,
'statsmodels': self._run_multitest_statsmodels
}
self.pop = set(pop)
self.pop_n = len(pop)
self.assoc = assoc
self.obo_dag = obo_dag
self.alpha = alpha
if methods is None:
methods = ["bonferroni", "sidak", "holm"]
self.methods = Methods(methods)
self.pval_obj = FisherFactory(**kws).pval_obj
if propagate_counts:
update_association(assoc, obo_dag, kws.get('relationships', None))
## BROAD broad_goids = get_goids_to_remove(kws.get('remove_goids'))
## BROAD if broad_goids:
## BROAD assoc = self._remove_assc_goids(assoc, broad_goids)
self.go2popitems = get_terms("population", pop, assoc, obo_dag,
self.log)
def __init__(self,
pop,
assoc,
obo_dag,
propagate_counts=True,
alpha=.05,
methods=None,
**kws):
self.log = kws['log'] if 'log' in kws else sys.stdout
self._run_multitest = {
'local': lambda iargs: self._run_multitest_local(iargs),
'statsmodels': lambda iargs: self._run_multitest_statsmodels(iargs)
}
self.pop = pop
self.pop_n = len(pop)
self.assoc = assoc
self.obo_dag = obo_dag
self.alpha = alpha
if methods is None:
methods = ["bonferroni", "sidak", "holm"]
self.methods = Methods(methods)
self.pval_obj = FisherFactory(**kws).pval_obj
if propagate_counts:
sys.stderr.write("Propagating term counts to parents ..\n")
obo_dag.update_association(assoc)
self.go2popitems = get_terms("population", pop, assoc, obo_dag,
self.log)
def run_study(self, study, **kws):
"""Run Gene Ontology Enrichment Study (GOEA) on study ids."""
study_name = kws.get('name', 'current')
log = self._get_log_or_prt(kws)
if log:
log.write(
'\nRun {OBJNAME} Gene Ontology Analysis: {STU} study set of {N} IDs ...'
.format(OBJNAME=self.name, N=len(study), STU=study_name))
if not study:
return []
# Key-word arguments:
methods = Methods(kws['methods']) if 'methods' in kws else self.methods
alpha = kws['alpha'] if 'alpha' in kws else self.alpha
# Calculate uncorrected pvalues
results = self.get_pval_uncorr(study, log)
if not results:
return []
if log is not None:
log.write(" {MSG}\n".format(
MSG="\n ".join(self.get_results_msg(results, study))))
# Do multipletest corrections on uncorrected pvalues and update results
self._run_multitest_corr(results, methods, alpha, study, log)
for rec in results:
# get go term for name and level
rec.set_goterm(self.obo_dag)
# 'keep_if' can be used to keep only significant GO terms. Example:
# >>> keep_if = lambda nt: nt.p_fdr_bh < 0.05 # if results are significant
# >>> goea_results = goeaobj.run_study(geneids_study, keep_if=keep_if)
if 'keep_if' in kws:
keep_if = kws['keep_if']
results = [r for r in results if keep_if(r)]
# Default sort order:
results.sort(key=lambda r: [r.enrichment, r.NS, r.p_uncorrected])
return results # list of GOEnrichmentRecord objects
def test_init_methods():
"""Test initializing methods."""
mobj = Methods()
assert mobj._srcmethod2fieldname == get_exp_fieldnames()
assert mobj.getmsg_valid_methods() == get_expstr_fieldnames()
assert mobj.methods == [mobj.NtMethodInfo(source='local', method='bonferroni', fieldname='bonferroni')]
mobj._add_method_src('statsmodels', 'fdr_bh')
assert mobj.methods == [
mobj.NtMethodInfo(source='local', method='bonferroni', fieldname='bonferroni'),
mobj.NtMethodInfo(source='statsmodels', method='fdr_bh', fieldname='fdr_bh')]
sm_methods = ['sm_{}'.format(m) for m in mobj.all_methods[1][1]] # statsmodels
mobj._init_methods(sm_methods)
assert mobj.methods == [
mobj.NtMethodInfo(source='statsmodels', method='bonferroni', fieldname='sm_bonferroni'),
mobj.NtMethodInfo(source='statsmodels', method='sidak', fieldname='sm_sidak'),
mobj.NtMethodInfo(source='statsmodels', method='holm-sidak', fieldname='sm_holm-sidak'),
mobj.NtMethodInfo(source='statsmodels', method='holm', fieldname='sm_holm'),
mobj.NtMethodInfo(source='statsmodels', method='simes-hochberg', fieldname='sm_simes-hochberg'),
mobj.NtMethodInfo(source='statsmodels', method='hommel', fieldname='sm_hommel'),
mobj.NtMethodInfo(source='statsmodels', method='fdr_bh', fieldname='sm_fdr_bh'),
mobj.NtMethodInfo(source='statsmodels', method='fdr_by', fieldname='sm_fdr_by'),
mobj.NtMethodInfo(source='statsmodels', method='fdr_tsbh', fieldname='sm_fdr_tsbh'),
mobj.NtMethodInfo(source='statsmodels', method='fdr_tsbky', fieldname='sm_fdr_tsbky'),
mobj.NtMethodInfo(source='statsmodels', method='fdr_gbs', fieldname='sm_fdr_gbs')]
def _init_args(self):
"""Get enrichment arg parser."""
#pylint: disable=invalid-name
p = argparse.ArgumentParser(
__doc__, formatter_class=argparse.ArgumentDefaultsHelpFormatter)
p.add_argument('filenames',
type=str,
nargs=3,
help='data/study data/population data/association')
p.add_argument(
'--annofmt',
default=None,
type=str,
help=('Annotation file format. '
'Not needed if type can be determined using filename'),
choices=['gene2go', 'gaf', 'gpad', 'id2gos'])
p.add_argument(
'--taxid',
default=9606,
type=int,
help=
"When using NCBI's gene2go annotation file, specify desired taxid")
p.add_argument('--alpha',
default=0.05,
type=float,
help='Test-wise alpha for multiple testing')
p.add_argument(
'--pval',
default=.05,
type=float,
help='Only print results with uncorrected p-value < PVAL.')
p.add_argument('--pval_field',
type=str,
help='Only print results when PVAL_FIELD < PVAL.')
p.add_argument('--outfile',
default=None,
type=str,
help='Write enrichment results into xlsx or tsv file')
p.add_argument('--ns',
default='BP,MF,CC',
type=str,
help='Limit GOEA to specified branch categories. '
'BP=Biological Process; '
'MF=Molecular Function; '
'CC=Cellular Component')
p.add_argument(
'--id2sym',
default=None,
type=str,
help='ASCII file containing one geneid and its symbol per line')
p.add_argument(
'--sections',
default=None,
type=str,
help=('Use sections file for printing grouped GOEA results. '
'Example SECTIONS values:\n'
'goatools.test_data.sections.gjoneska_pfenning \n'
'goatools/test_data/sections/gjoneska_pfenning.py \n'
'data/gjoneska_pfenning/sections_in.txt\n'))
p.add_argument(
'--outfile_detail',
type=str,
help=(
'Write enrichment results into a text file \n'
'containing the following information: \n'
'1) GOEA GO terms, grouped into sections \n\n'
'2) List of genes and ASCII art showing section membership \n'
'3) Detailed list of each gene and GO terms w/their P-values \n'
))
p.add_argument(
'--compare',
dest='compare',
default=False,
action='store_true',
help="the population file as a comparison group. if this "
"flag is specified, the population is used as the study "
"plus the `population/comparison`")
p.add_argument(
'--ratio',
dest='ratio',
type=float,
default=None,
help="only show values where the difference between study "
"and population ratios is greater than this. useful for "
"excluding GO categories with small differences, but "
"containing large numbers of genes. should be a value "
"between 1 and 2. ")
p.add_argument('--indent',
dest='indent',
default=False,
action='store_true',
help="indent GO terms")
p.add_argument('--obo',
default="go-basic.obo",
type=str,
help="Specifies location and name of the obo file")
p.add_argument('--no_propagate_counts',
default=False,
action='store_true',
help="Do not propagate counts to parent terms")
p.add_argument('--method',
default="bonferroni,sidak,holm,fdr_bh",
type=str,
help=Methods().getmsg_valid_methods())
p.add_argument('--pvalcalc',
default="fisher",
type=str,
help=str(FisherFactory()))
p.add_argument(
'--min_overlap',
default=0.7,
type=float,
help=
"Check that a minimum amount of study genes are in the population")
p.add_argument('--goslim',
default='goslim_generic.obo',
type=str,
help="The GO slim file is used when grouping GO terms.")
p.add_argument(
'--ev_inc',
type=str,
help=
"Include specified evidence codes and groups separated by commas")
p.add_argument(
'--ev_exc',
type=str,
help=
"Exclude specified evidence codes and groups separated by commas")
p.add_argument('--ev_help',
dest='ev_help',
action='store_false',
help="Print all Evidence codes, with descriptions")
p.add_argument('--ev_help_short',
dest='ev_help_short',
action='store_false',
help="Print all Evidence codes")
if len(sys.argv) == 1:
sys.exit(not p.print_help())
self._prt_evidence_codes(set(sys.argv[1:]))
args = p.parse_args() # Namespace object from argparse
self._check_input_files(args, p)
return args
"containing large numbers of genes. should be a value "
"between 1 and 2. ")
p.add_argument('--fdr', dest='fdr', default=False,
action='store_true',
help="Calculate the false discovery rate (alt. to the "
"Bonferroni but slower)")
p.add_argument('--indent', dest='indent', default=False,
action='store_true', help="indent GO terms")
p.add_argument('--obo', default="go-basic.obo", type=str,
help="Specifies location and name of the obo file")
p.add_argument('--no_propagate_counts', default=False, action='store_true',
help="Do not propagate counts to parent terms")
p.add_argument('--outfile', default=None, type=str,
help="Write enrichment results into xlsx or tsv file")
p.add_argument('--method', default="bonferroni,sidak,holm", type=str,
help=Methods().getmsg_valid_methods())
args = p.parse_args()
check_input_files(args, p)
min_ratio = args.ratio
if min_ratio is not None:
assert 1 <= min_ratio <= 2
study_fn, pop_fn, assoc_fn = args.filenames
study, pop = read_geneset(study_fn, pop_fn, compare=args.compare)
print("Study: {0} vs. Population {1}".format(len(study), len(pop)), file=sys.stderr)
if not args.compare: # sanity check
if len(pop) < len(study):
exit("\nERROR: The study file contains more elements than the population file. "
def _init_args(self):
"""Get enrichment arg parser."""
#pylint: disable=invalid-name
p = argparse.ArgumentParser(
__doc__, formatter_class=argparse.ArgumentDefaultsHelpFormatter)
p.add_argument('filenames',
type=str,
nargs=3,
help='data/study data/population data/association')
p.add_argument('--alpha',
default=0.05,
type=float,
help='Test-wise alpha for multiple testing')
p.add_argument(
'--pval',
default=.05,
type=float,
help='Only print results with uncorrected p-value < PVAL.')
p.add_argument('--pval_field',
type=str,
help='Only print results when PVAL_FIELD < PVAL.')
p.add_argument('--outfile',
default=None,
type=str,
help='Write enrichment results into xlsx or tsv file')
p.add_argument(
'--sections',
default=None,
type=str,
help=('Use sections file for printing grouped GOEA results. '
'Example SECTIONS values:\n'
'goatools.test_data.sections.gjoneska_pfenning \n'
'goatools/test_data/sections/gjoneska_pfenning.py \n'
'data/gjoneska_pfenning/sections_in.txt\n'))
p.add_argument(
'--outfile_detail',
type=str,
help=(
'Write enrichment results into a text file \n'
'containing the following information: \n'
'1) GOEA GO terms, grouped into sections \n\n'
'2) List of genes and ASCII art showing section membership \n'
'3) Detailed list of each gene and GO terms w/their P-values \n'
))
p.add_argument(
'--compare',
dest='compare',
default=False,
action='store_true',
help="the population file as a comparison group. if this "
"flag is specified, the population is used as the study "
"plus the `population/comparison`")
p.add_argument(
'--ratio',
dest='ratio',
type=float,
default=None,
help="only show values where the difference between study "
"and population ratios is greater than this. useful for "
"excluding GO categories with small differences, but "
"containing large numbers of genes. should be a value "
"between 1 and 2. ")
p.add_argument('--indent',
dest='indent',
default=False,
action='store_true',
help="indent GO terms")
p.add_argument('--obo',
default="go-basic.obo",
type=str,
help="Specifies location and name of the obo file")
p.add_argument('--no_propagate_counts',
default=False,
action='store_true',
help="Do not propagate counts to parent terms")
p.add_argument('--method',
default="bonferroni,sidak,holm,fdr_bh",
type=str,
help=Methods().getmsg_valid_methods())
p.add_argument('--pvalcalc',
default="fisher",
type=str,
help=str(FisherFactory()))
p.add_argument(
'--min_overlap',
default=0.7,
type=float,
help=
"Check that a minimum amount of study genes are in the population")
p.add_argument('--goslim',
default='goslim_generic.obo',
type=str,
help="The GO slim file is used when grouping GO terms.")
if len(sys.argv) == 1:
sys.exit(not p.print_help())
args = p.parse_args() # Namespace object from argparse
self._check_input_files(args, p)
return args
def get_arg_parser():
"""Get enrichment arg parser."""
#pylint: disable=invalid-name
p = argparse.ArgumentParser(
__doc__, formatter_class=argparse.ArgumentDefaultsHelpFormatter)
p.add_argument('filenames',
type=str,
nargs=3,
help='data/study data/population data/association')
p.add_argument('--alpha',
default=0.05,
type=float,
help="Test-wise alpha for multiple testing ")
p.add_argument(
'--pval',
default=.05,
type=float,
help="Only print out when uncorrected p-value < this value.")
p.add_argument('--compare',
dest='compare',
default=False,
action='store_true',
help="the population file as a comparison group. if this "
"flag is specified, the population is used as the study "
"plus the `population/comparison`")
p.add_argument('--ratio',
dest='ratio',
type=float,
default=None,
help="only show values where the difference between study "
"and population ratios is greater than this. useful for "
"excluding GO categories with small differences, but "
"containing large numbers of genes. should be a value "
"between 1 and 2. ")
p.add_argument('--indent',
dest='indent',
default=False,
action='store_true',
help="indent GO terms")
p.add_argument('--obo',
default="go-basic.obo",
type=str,
help="Specifies location and name of the obo file")
p.add_argument('--no_propagate_counts',
default=False,
action='store_true',
help="Do not propagate counts to parent terms")
p.add_argument('--outfile',
default=None,
type=str,
help="Write enrichment results into xlsx or tsv file")
p.add_argument('--method',
default="bonferroni,sidak,holm,fdr_bh",
type=str,
help=Methods().getmsg_valid_methods())
p.add_argument('--pvalcalc',
default="fisher",
type=str,
help=str(FisherFactory()))
p.add_argument(
'--min_overlap',
default=0.7,
type=float,
help="Check that a minimum amount of study genes are in the population"
)
if len(sys.argv) == 1:
sys.exit(not p.print_help())
args = p.parse_args() # Namespace object from argparse
_check_input_files(args, p)
return args
help=('Write enrichment results into a text file \n'
'containing the following information: \n'
'1) GOEA GO terms, grouped into sections \n\n'
'2) List of genes and ASCII art showing section membership \n'
'3) Detailed list of each gene and GO terms w/their P-values \n'),
abbrev='D')
@plac.opt('ratio', type=float,
help="only show values where the difference between study "
"and population ratios is greater than this. useful for "
"excluding GO categories with small differences, but "
"containing large numbers of genes. should be a value "
"between 1 and 2. ")
@plac.opt('relationships', abbrev='R',
help=('Propagate counts up user-specified relationships ( comma separated ), which include: '
'{RELS}').format(RELS=' '.join(RELATIONSHIP_SET)))
@plac.opt('method', type=str, help=Methods().getmsg_valid_methods())
@plac.opt('pvalcalc', type=str, help=str(FisherFactory()), abbrev='calc')
@plac.opt('min_overlap', type=float,
help="Check that a minimum amount of study genes are in the population",
abbrev='M')
@plac.opt('goslim', type=str, help="The GO slim file is used when grouping GO terms.")
def run(name='human', taxid=9606, download=False,
alpha=0.05, pval=.05, field='p_uncorrected', outfile='result.tsv',
ns='BP,MF,CC', id2sym=None, detail='', sections=None,
compare=False, ratio=None, prtstd=False, indent=False,
noprop=False, relationship=False, relationships='', plot=False,
enrich=False, method="bonferroni,sidak,holm,fdr_bh", pvalcalc="fisher",
min_overlap=0.7, goslim='goslim_generic.obo', inc='', exc='',
*study):
# Construct arguments to pass down to GO.
go_params = dict(alpha=alpha, pval=pval, pval_field=field, outfile=outfile,
