def do_go_analysis(go_data,
genes,
p_value_threshold,
method='weight',
elim_cut_off=0.01):
"""
Do a topGO analysis on the genes.
@arg method: Can be 'classic', 'weight' or 'elim'
"""
# update the go data with the given genes
go_data = update_go_data(go_data, genes)
# get the significant groups
if 'weight' == method:
test_stat = r.new("weightCount",
testStatistic=r.GOFisherTest,
name="Fisher test",
sigRatio="ratio")
elif 'classic' == method:
test_stat = r.new("classicCount",
testStatistic=r.GOFisherTest,
name="Fisher test")
elif 'elim' == method:
test_stat = r.new("elimCount",
testStatistic=r.GOFisherTest,
name="Fisher test",
cutOff=elim_cut_off)
else:
raise ValueError('%s: Unknown topGO method' % method)
sig_groups = r.getSigGroups(go_data, test_stat)
args = {method: sig_groups}
results_unthresholded = r.GenTable(
go_data,
#ranksOf="classic",
orderBy=method,
**args)
# only keep those results above the threshold
assert method == results_unthresholded.colnames()[
5] # make sure looking at correct column
# which rows are above the threshold?
passed = rpy2.robjects.BoolVector([
p_value_from_r(p) <= p_value_threshold
for p in results_unthresholded[5]
])
return r.subset(results_unthresholded, passed)
def do_go_analysis(go_data, genes, p_value_threshold, method='weight', elim_cut_off=0.01):
"""
Do a topGO analysis on the genes.
@arg method: Can be 'classic', 'weight' or 'elim'
"""
# update the go data with the given genes
go_data = update_go_data(go_data, genes)
# get the significant groups
if 'weight' == method:
test_stat = r.new(
"weightCount",
testStatistic=r.GOFisherTest,
name="Fisher test",
sigRatio="ratio"
)
elif 'classic' == method:
test_stat = r.new(
"classicCount",
testStatistic=r.GOFisherTest,
name="Fisher test"
)
elif 'elim' == method:
test_stat = r.new(
"elimCount",
testStatistic=r.GOFisherTest,
name="Fisher test",
cutOff = elim_cut_off
)
else:
raise ValueError('%s: Unknown topGO method' % method)
sig_groups = r.getSigGroups(go_data, test_stat)
args = {method : sig_groups}
results_unthresholded = r.GenTable(
go_data,
#ranksOf="classic",
orderBy=method,
**args
)
# only keep those results above the threshold
assert method == results_unthresholded.colnames()[5] # make sure looking at correct column
# which rows are above the threshold?
passed = rpy2.robjects.BoolVector([p_value_from_r(p) <= p_value_threshold for p in results_unthresholded[5]])
return r.subset(results_unthresholded, passed)
def go_enrichment(gene_association, subset, algo, minnode):
significant = []
geneID2GO = R.readMappings(file=gene_association)
refset = R.names(geneID2GO)
testset = R.scan(file=subset, what=R.character())
# Double %% escapes % in strings1
genes_of_interest = R("factor(as.integer(%s %%in%% %s))" % (refset.r_repr(), testset.r_repr()))
# Use the setNames function instead of this R code,
# names(genes_of_interest) <- refset
# because Python cannot assign to values
genes_of_interest = R.setNames(genes_of_interest, refset)
for o in ["MF", "BP", "CC"]:
GOdata = R.new(
"topGOdata",
ontology=o,
allGenes=genes_of_interest,
annot=R["annFUN.gene2GO"],
gene2GO=geneID2GO,
nodeSize=minnode,
)
test = R.runTest(GOdata, algorithm=algo, statistic="fisher")
pvalues = list(R.score(test))
terms = list(R.names(R.score(test)))
terms, pvalues, padjusted = adjust_pvalues(terms, pvalues)
for i, t in enumerate(terms):
if pvalues[i] < 0.05:
significant.append([t, str(pvalues[i]), str(padjusted[i])])
return significant
def go_enrichment(genes2go, genes_list, algo, nodeSize):
init_topGO()
#init_qvalue()
genes2go_map = R.readMappings(file=genes2go)
subset = genes_list
refset = R.names(genes2go_map)
genes_of_interest = R("factor(as.integer(%s %%in%% %s))" % (refset.r_repr(), subset.r_repr()))
genes_of_interest = R.setNames(genes_of_interest, refset)
score = collections.defaultdict(dict)
for o in ["MF", "BP", "CC"]:
#for o in ["MF"]:
GOdata = R.new("topGOdata",
ontology=o,
annot=R["annFUN.gene2GO"],
allGenes=genes_of_interest,
gene2GO=genes2go_map,
nodeSize=nodeSize
)
scoreR = R.score(R.runTest(GOdata, algorithm=algo, statistic="fisher"))
for i in range(len(scoreR)):
if scoreR[i] < 0.05:
score[scoreR.names[i]] = {"pval": scoreR[i]}
score = collections.OrderedDict(sorted(score.items(), key=lambda t: t[1]))
score = __add_adjusted_pvalues(score)
score = __add_GO_info(score)
__save_results(score)