def get_sequences(version):
filters = {}
filters['page_index'], filters['page_size'] = _get_page_arg()
filters['keyword'] = _get_keyword_arg()
_get_offset_arg(filters)
_get_lrr_count_arg(filters)
_get_species_arg(filters)
filters['page_index'] = filters['page_index'] * filters['page_size']
logging.debug(str.format("Filters: {}", filters))
seqs, total = dao.query_sequences(filters, get_version_arg(version))
# find motifs and nsites
seq_ids = [seq.seq_id for seq in seqs]
seq_ids_to_motifs = dao.find_motifs_by_seq_ids(seq_ids,
get_version_arg(version))
seq_ids_to_nsites = dao.find_nsites_by_seq_ids(seq_ids)
# find tags, the overlap mark is tagged on version 1 motifs
if get_version_arg(version) == 1:
ids_to_tag_names = dao.find_tags_by_motif_ids(
set([
m.id for motifs in seq_ids_to_motifs.values() for m in motifs
]))
else:
ids_to_tag_names = {}
result = {'sequences': [], 'total': total}
for seq in seqs:
result['sequences'].append(
sequence_entity_to_output(seq,
seq_ids_to_motifs.get(seq.seq_id, []),
seq_ids_to_nsites.get(seq.seq_id, []),
ids_to_tag_names))
return response_ok(result, True)
def main():
with dao.query_session() as session:
seqs = dao.sequence.find_all_seqs(session)
step1(seqs)
logging.info("Step 1 end")
# since only five SGs were considered from step 3.4, the seq was redefined
relevant_seqs = [seq for seq in seqs if seq.subgroup in SUBGROUPS]
relevant_seq_ids = [seq.seq_id for seq in relevant_seqs]
seq_ids_to_nsites = dao.find_nsites_by_seq_ids(relevant_seq_ids)
seq_ids_to_lrrs = dao.find_motifs_by_seq_ids(relevant_seq_ids,
MOTIF_VERSION,
with_wrong=False)
step3_5(relevant_seqs, seq_ids_to_nsites, seq_ids_to_lrrs)
logging.info("Step 3.5 end")
step3_7(relevant_seqs, seq_ids_to_nsites, seq_ids_to_lrrs)
logging.info("Step 3.7 end")
step3_9(relevant_seqs, seq_ids_to_nsites, seq_ids_to_lrrs)
logging.info("Step 3.9 end")
step3_10(seq_ids_to_lrrs, seqs)
logging.info("Step 3.10 end")
step3_11(seqs, seq_ids_to_lrrs)
logging.info("Step 3.11 end")
full_seq_ids = [seq.seq_id for seq in seqs]
full_seq_ids_to_nsites = dao.find_nsites_by_seq_ids(full_seq_ids)
full_seq_ids_to_lrrs = dao.find_motifs_by_seq_ids(full_seq_ids,
MOTIF_VERSION,
with_wrong=False)
supplement_for_review(seqs, full_seq_ids_to_nsites, full_seq_ids_to_lrrs)
logging.info("Step supplement for review end")
def main():
with dao.query_session() as session:
seqs = dao.sequence.find_all_seqs(session)
step1(seqs)
logging.info("Step 1 end")
# 3.4开始，只关注特定的5个亚家族，因此对seq做精简
relevant_seqs = [seq for seq in seqs if seq.subgroup in SUBGROUPS]
relevant_seq_ids = [seq.seq_id for seq in relevant_seqs]
seq_ids_to_nsites = dao.find_nsites_by_seq_ids(relevant_seq_ids)
seq_ids_to_lrrs = dao.find_motifs_by_seq_ids(relevant_seq_ids,
MOTIF_VERSION,
with_wrong=False)
step3_5(relevant_seqs, seq_ids_to_nsites, seq_ids_to_lrrs)
logging.info("Step 3.5 end")
step3_7(relevant_seqs, seq_ids_to_nsites, seq_ids_to_lrrs)
logging.info("Step 3.7 end")
step3_9(relevant_seqs, seq_ids_to_nsites, seq_ids_to_lrrs)
logging.info("Step 3.9 end")
step3_10(seq_ids_to_lrrs, seqs)
logging.info("Step 3.10 end")
step3_11(seqs, seq_ids_to_lrrs)
logging.info("Step 3.11 end")
full_seq_ids = [seq.seq_id for seq in seqs]
full_seq_ids_to_nsites = dao.find_nsites_by_seq_ids(full_seq_ids)
full_seq_ids_to_lrrs = dao.find_motifs_by_seq_ids(full_seq_ids,
MOTIF_VERSION,
with_wrong=False)
supplement_for_review(seqs, full_seq_ids_to_nsites, full_seq_ids_to_lrrs)
logging.info("Step supplement for review end")
def step3_11(seqs, seq_ids_to_lrrs):
seq_ids = set([seq.seq_id for seq in seqs])
seq_ids_to_nsites = dao.find_nsites_by_seq_ids(seq_ids)
subgroups_to_ntypes_to_count = {}
for seq in seqs:
subgroup = _get_subgroup(seq, seq_ids_to_lrrs)
if subgroup is None:
continue
if subgroup not in subgroups_to_ntypes_to_count:
subgroups_to_ntypes_to_count[subgroup] = {'S': 0, 'T': 0}
nsites = seq_ids_to_nsites.get(seq.seq_id, [])
for nsite in nsites:
if nsite.ntype == dao.nsite.S:
subgroups_to_ntypes_to_count[subgroup]['S'] += 1
else:
subgroups_to_ntypes_to_count[subgroup]['T'] += 1
generate_histogram(subgroups_to_ntypes_to_count, 'ntype',
'step3_11_n_count_by_subgroup_type.png')