def query_single_gset(query, all_genes=[], gene_sets=[]):
res = calc_pvalues(query, gene_sets, background=all_genes)
if len(res) == 0:
return 0, None
else:
set_names, p_vals, overlap_size, gset_size, overlapped_genes = res
# by default Benjamini-Hochberg
q_vals, rejs = multiple_testing_correction(p_vals)
#min_log_q_vals.append(np.log10(min(q_vals)))
best_hit_ndx = np.argmin(q_vals)
return -np.log10(q_vals[best_hit_ndx]), set_names[best_hit_ndx]
def enrich(self, gmt):
"""use local mode
p = p-value computed using the Fisher exact test (Hypergeometric test)
Not implemented here:
combine score = log(p)·z
see here: http://amp.pharm.mssm.edu/Enrichr/help#background&q=4
columns contain:
Term Overlap P-value Adjusted_P-value Genes
"""
if isscalar(self.background):
if isinstance(self.background, int) or self.background.isdigit():
self._bg = int(self.background)
elif isinstance(self.background, str):
# self.background = set(reduce(lambda x,y: x+y, gmt.values(),[]))
self._bg = self.get_background()
self._logger.info("Background: found %s genes" %
(len(self._bg)))
else:
raise Exception("Unsupported background data type")
else:
# handle array object: nd.array, list, tuple, set, Series
try:
it = iter(self.background)
self._bg = set(self.background)
except TypeError:
self._logger.error("Unsupported background data type")
# statistical testing
hgtest = list(
calc_pvalues(query=self._gls, gene_sets=gmt, background=self._bg))
if len(hgtest) > 0:
terms, pvals, olsz, gsetsz, genes = hgtest
fdrs, rej = multiple_testing_correction(
ps=pvals, alpha=self.cutoff, method='benjamini-hochberg')
# save to a dataframe
odict = OrderedDict()
odict['Term'] = terms
odict['Overlap'] = list(
map(lambda h, g: "%s/%s" % (h, g), olsz, gsetsz))
odict['P-value'] = pvals
odict['Adjusted P-value'] = fdrs
# odict['Reject (FDR< %s)'%self.cutoff ] = rej
odict['Genes'] = [";".join(g) for g in genes]
res = pd.DataFrame(odict)
return res
return
def enrich(self, gmt):
"""use local mode
p = p-value computed using the Fisher exact test (Hypergeometric test)
Not implemented here:
combine score = log(p)·z
see here: http://amp.pharm.mssm.edu/Enrichr/help#background&q=4
columns contain:
Term Overlap P-value Adjusted_P-value Genes
"""
if isinstance(self.background, str):
# self.background = set(reduce(lambda x,y: x+y, gmt.values(),[]))
df = self.get_background()
# input id type: entrez or gene_name
if self._isezid:
bg = df['entrezgene'].astype(int)
else:
bg = df['external_gene_name']
self._bg = set(bg.unique())
self._logger.warning("Background: %s %s genes with GO_IDs. " %
(self._bg, self.background))
self._logger.warning(
"If this is not you wanted, please give a number to background argument"
)
hgtest = list(
calc_pvalues(query=self._gls, gene_sets=gmt, background=self._bg))
if len(hgtest) > 0:
terms, pvals, olsz, gsetsz, genes = hgtest
fdrs, rej = multiple_testing_correction(
ps=pvals, alpha=self.cutoff, method='benjamini-hochberg')
# save to a dataframe
odict = OrderedDict()
odict['Term'] = terms
odict['Overlap'] = list(
map(lambda h, g: "%s/%s" % (h, g), olsz, gsetsz))
odict['P-value'] = pvals
odict['Adjusted P-value'] = fdrs
# odict['Reject (FDR< %s)'%self.cutoff ] = rej
odict['Genes'] = [";".join(g) for g in genes]
res = pd.DataFrame(odict)
return res
return
def query_single_gset(query, all_genes=[], gene_sets=[]):
res = calc_pvalues(query, gene_sets, background=all_genes)
if len(res) == 0:
return 0, None,0
else:
set_names, p_vals, overlap_size, gset_size, overlapped_genes = res
# by default Benjamini-Hochberg
q_vals, rejs = multiple_testing_correction(p_vals)
#min_log_q_vals.append(np.log10(min(q_vals)))
best_fc = 0
best_hit_ndx = -1
for i in range(0,len(q_vals)):
if q_vals[i]<0.05:
obs = float(overlap_size[i])
expec = float(gset_size[i])*len(query)/len(all_genes)
fc = obs/expec
if fc > best_fc:
best_fc=fc
best_hit_ndx = i
if best_hit_ndx == -1:
return 0, None,0
else:
return -np.log10(q_vals[best_hit_ndx]), set_names[best_hit_ndx], np.log10(best_fc)
