def test_replace_tags_by_motifs(self):
seq_ids_to_motifs = {'SEQ1': [], 'SEQ2': []}
with query_session() as session:
seq_ids_to_motifs['SEQ1'] = dao.motif.find_motifs_by_seq_id(
session, 'SEQ1', MOTIF_VERSION)
seq_ids_to_motifs['SEQ2'] = dao.motif.find_motifs_by_seq_id(
session, 'SEQ2', MOTIF_VERSION)
mids_to_tags = {seq_ids_to_motifs['SEQ1'][0].id: ['a', 'b', 'c']}
cls = dao.motif.get_entity(MOTIF_VERSION)
with query_session() as session:
motifs = session.query(cls).filter(cls.id.in_(
mids_to_tags.keys())).all()
self.assertEqual(1, len(motifs))
self.assertTrue(motifs[0].correct)
dao.replace_tags_by_motifs(mids_to_tags, MOTIF_VERSION)
seq = dao.find_seq_by_id('SEQ1')
self.assertIsNotNone(seq)
self.assertEqual(
10,
len(
dao.find_correct_motifs_by_seq_ids([
'SEQ1',
], MOTIF_VERSION)['SEQ1']))
with query_session() as session:
motifs = session.query(cls).filter(cls.id.in_(
mids_to_tags.keys())).all()
self.assertEqual(1, len(motifs))
self.assertTrue(motifs[0].correct)
def query_density(seq_id):
with dao.query_session() as session:
motifs = dao.motif.find_motifs_by_seq_id(session,
seq_id,
MOTIF_VERSION,
with_wrong=False)
seq = dao.find_seq_by_id(seq_id)
return motifs, seq
def obtain_value(seq_id):
with dao.query_session() as session:
motifs = dao.motif.find_motifs_by_seq_id(session,
seq_id,
MOTIF_VERSION,
with_wrong=False)
nsites = dao.nsite.find_nsites_by_seq_id(session, seq_id)
seq = dao.find_seq_by_id(seq_id)
return motifs, nsites, seq
def test_add_seq(self):
seq = dao.find_seq_by_id('SEQ1')
self.assertIsNotNone(seq)
self.assertEqual(
10,
len(
dao.find_correct_motifs_by_seq_ids([
'SEQ1',
], MOTIF_VERSION)['SEQ1']))
self.assertEqual('SEQABCDEFG', seq.seq)
def read_in_scratch_result(result):
ss = SeqIO.read(result.ss, 'fasta')
ss8 = SeqIO.read(result.ss8, 'fasta')
acc = SeqIO.read(result.acc, 'fasta')
acc20 = read_in_acc20(result.acc20)
if acc20 is None:
return
if ss.id != ss8.id or ss.id != acc.id or ss.id != acc20.id:
logging.error(
str.format(
"The result for seq {} is invalid, the seq id is different",
ss.id))
return
if len(ss.seq) != len(ss8.seq) or len(ss.seq) != len(acc.seq) or len(
ss.seq) != len(acc20.seq):
logging.error(
str.format(
"The result for seq {} is invalid, the seq length is different",
ss.id))
return
seq_entity = None
try:
seq_entity = dao.find_seq_by_id(ss.id)
except Exception as e:
logging.error(
str.format(
"Failed to save result for seq {}, can not find the record in the db",
ss.id))
return
if seq_entity is None:
logging.error(
str.format(
"Failed to save result for seq {}, can not find the record in the db",
ss.id))
return
if len(ss.seq) != len(seq_entity.seq):
logging.error(
str.format(
"The result for seq {} is invalid, the seq length is different with db",
ss.id))
return
logging.info("Update scratch result for seq " + ss.id)
dao.update_seq_for_scratch_result(ss.id, str(ss.seq), str(ss8.seq),
str(acc.seq), ' '.join(acc20.seq))
def test_unmark_wrong_with_other_wrong(self):
# unmark一个wrong motif的时候，考虑已存在的mark为wrong的motif，不应该影响到正常的计算
set_up_seq('SEQ4', 'SEQ4ABCDABCDABCD', start=10, count=2, step=10)
seq4 = dao.find_seq_by_id('SEQ4')
motifs = dao.find_correct_motifs_by_seq_ids([
'SEQ4',
], MOTIF_VERSION)['SEQ4']
self.assertEqual(2, len(motifs))
mark_wrong(seq4.id, motifs[1].id)
mark_wrong(seq4.id, motifs[0].id)
self.assertEqual(
0,
len(
dao.find_correct_motifs_by_seq_ids([
'SEQ4',
], MOTIF_VERSION)['SEQ4']))
result = unmark_wrong(seq4.id, motifs[0].id)
self.assertEqual(
1,
len(
dao.find_correct_motifs_by_seq_ids([
'SEQ4',
], MOTIF_VERSION)['SEQ4']))
def obtain_value(seq_id):
with dao.query_session() as session:
nsites = dao.nsite.find_nsites_by_seq_id(session, seq_id)
seq = dao.find_seq_by_id(seq_id)
return nsites, seq
import dao
from tools import motifs as motif_tool, pssm_matrix
MOTIF_VERSION = 2
with dao.session_scope() as session:
matrix = pssm_matrix.calc_pssm_matrix(dao.find_baseline_motifs())
logging.info(matrix.pssm)
cls = dao.motif.get_entity(MOTIF_VERSION)
motifs = session.query(cls).filter(cls.manually_add).all()
results = []
for m in motifs:
seq = dao.find_seq_by_id(m.seq_id)
motif_seq = seq.seq[m.offset:m.offset + 24]
score = motif_tool.calc_pssm_score(motif_seq[:16], matrix)
m.score = score
results.append({
'seq_id': m.seq_id,
'offset': m.offset,
'seq': motif_seq,
'score': score
})
logging.info(str.format("Total manually records {}", len(results)))
results.sort(key=lambda k: k['score'])
for r in results:
logging.info(
str.format(