class GoEnrich():
def __init__(self):
obodag = GODag("../Data/evaluation_reference/goslim_yeast.obo")
background = [line.strip() for line in open('../Data/evaluation_reference/gene_list.txt')]
geneid2gos_yeast = read_associations('../Data/evaluation_reference/geneid2gos_yeast.txt')
self.goeaobj = GOEnrichmentStudy(
background,
geneid2gos_yeast,
obodag,
propogate_counts=False,
alpha=0.05,
methods=['fdr_bh'])
def measure_enrichment(self,
gene_set=['YML106W', 'YKL135C', 'YDR516C',
'YLR420W', 'YNL111C', 'YHR007C',
'YLR014C', 'YKL216W', 'YNL078W',
'YJR005W', 'YJL130C'],
run_name='base',
cluster_id=1):
gene_ids = ['YML106W', 'YKL135C', 'YDR516C', 'YLR420W', 'YNL111C',
'YHR007C', 'YLR014C', 'YKL216W', 'YNL078W', 'YJR005W',
'YJL130C']
goea_results_all = self.goeaobj.run_study(gene_ids)
# we can get significant only
# goea_results_sig = [r for r in goea_results_all if r.p_fdr_bh < 0.05]
self.goeaobj.wr_txt("../Results/" + run_name + "_" + str(cluster_id) +
".txt", goea_results_all)
