def enrich_go(self):
"count ac2counts(local,global),sel_num"
sel_num = self.model.count()
go2ids = {'cc': {}, 'bp': {}, 'mf': {}}
m_mod = self.model.join(Go, on=(Contig.id == Go.contig))
for i in m_mod:
t2gos = json.loads(
url_tool.post(self.app_conf['url_go_extype'],
{'gos': i.gos[0].goid}))
for j, k in t2gos.items():
temp = [go2ids[j].setdefault(x, []).append(i.id) for x in k]
go2ids = {x: y for x, y in go2ids.items() if y}
# associated tree
total_gos = []
temp = [total_gos.extend(y.keys()) for x, y in go2ids.items()]
go_tree = json.loads(
url_tool.post(self.app_conf['url_go_type2tree'],
{'gos': ','.join(total_gos)}))
# global data input
t_info = json.load(
open(os.path.join(g.conf.static_dir, 'json', 'enrich_go.json')))
pool_info = {'num': t_info['pool_count'], 'go2num': {}}
for i, j in go2ids.items():
for t_goid in j.keys():
pool_info['go2num'][t_goid] = t_info['go2count'][i][t_goid]
return {
'sel_num': sel_num,
'go2ids': go2ids,
'pool_info': pool_info,
'go_tree': go_tree
}
def lineage(self):
"output taxid map to their lv name + taxname"
ids = [str(x.taxa_id) for x in self.model]
t_url = json.loads(url_tool.post(url_tool.make(g.conf.service_conf,'taxa_lv'),\
{'ids':','.join(ids)}))
return {
'tax2names': {
y: (t_url['lv_name'][x] + [t_url['taxname'][x]])
for x, y in enumerate(t_url['h_ids'])
},
'ex': t_url['ex'],
'lv_tag': t_url['default_lv']
}
def tax2comp(self):
"detect component group"
ids = [str(x.taxa_id) for x in self.model]
t_url = json.loads(
url_tool.post(self.app_conf['url_lineage_id'],
{'ids': ','.join(ids)}))
tax2comp = {
'33208': 'Coral',
'33630': 'Symbiodinium',
'33634': 'Symbiodinium',
'33090': 'Symbiodinium',
'2': 'Bacteria'
}
tax_set = set(tax2comp.keys())
res = {}
for i, j in t_url['tax2line'].items():
t_inter = tax_set.intersection(j)
if t_inter:
res[i] = tax2comp[t_inter.pop()]
else:
res[i] = 'intermediate homology'
return res
def enrich_kegg(self):
"count ac2counts(local,global),sel_num"
sel_num = self.model.count()
# get anno_db_id
o_db = Anno_db.get(Anno_db.name == 'kegg')
ac2ids = {}
m_mod = self.model.join(
Anno_info, on=(Contig.id == Anno_info.contig)).join(
Anno_hit, on=(Anno_hit.id == Anno_info.anno_hit)).join(
Anno_db, on=(Anno_db.id == Anno_hit.anno_db))
for i in m_mod.where(Anno_db.name == o_db.name):
for j in i.anno_infos:
if j.anno_hit.anno_db == o_db:
t_ac = j.anno_hit.ac
ac2ids.setdefault(t_ac, []).append(i.id)
# associated pathway
enrich_info = json.loads(
url_tool.post(self.app_conf['url_enrich_kegg'],
{'kos': ','.join(ac2ids.keys())}))
# global data input
t_info = json.load(
open(
os.path.join(g.conf.static_dir, g.conf.db_conn,
'enrich_kegg.json')))
enrich_info.update({
'path2count': {
x: t_info['path2count'][x]
for x in enrich_info['path2kos'] if x in t_info['path2count']
},
'count_pool': t_info['pool_count']
})
return {
'sel_num': sel_num,
'ac2ids': ac2ids,
'enrich_info': enrich_info
}