def load_all_sequences(filename, file_format):
start_time = time.time()
for seqRecord in SeqIO.parse(filename, file_format):
seq = seqRecord.seq.upper()
location = []
region = ''
for feature in seqRecord.features:
if feature.type == 'source':
start = feature.location.nofuzzy_start
end = feature.location.nofuzzy_end
country = feature.qualifiers.get('country', ['undefined: UND'])
location = [start, end]
region = country[0]
# count nucleoid stats if needed
# reset stats first
# count_nucleoid_stats(seq)
document = SequenceDocument(
version=str(seqRecord.id),
length=len(seq),
fasta=seqRecord.format('fasta'),
sequence=str(seq),
location=location,
region=region
)
document.save()
load_time = ExecutionTime(name="Sequence Load Time", seconds=time.time() - start_time)
load_time.save()
def parce_base_sequence(filePath, file_format):
for record in SeqIO.parse(filePath, file_format):
seqId = str(record.id)
seq = str(record.seq.upper())
document = SequenceDocument(
version=seqId,
length=len(seq),
fasta=record.format('fasta'),
sequence=seq,
name=record.name
)
document.save()
return document.id
def query_not_base_sequences():
length = 16569
cursor = SequenceDocument.objects(length__ne=length)
records = []
for r in cursor:
records.append(r)
del cursor
return records
def query_normal_length():
length = 377
cursor = SequenceDocument.objects(length=length)
records = []
for r in cursor:
records.append(r)
del cursor
return records
def save_stats_db(sequence):
rcrs_base_sequence = query_base_sequence('NC_012920')
rsrs_base_sequence = query_base_sequence('RSRS')
doc = SequenceDocument.objects(version=sequence.version)
rcrs_distance = count_distance(sequence.sequence, rcrs_base_sequence.sequence[16023:16400])
doc.update(distance_to_rCRS=rcrs_distance)
rsrs_distance = count_distance(sequence.sequence, rsrs_base_sequence.sequence[16023:16400])
doc.update(distance_to_RSRS=rsrs_distance)
def query_base_sequence(name):
return SequenceDocument.objects(name=name)[0]
def reset_sequence_documents_regions():
for seq_document in SequenceDocument.objects(length=377):
seq_document.delete()
def get_sequences_by_region(region):
return SequenceDocument.objects(region__contains=region).only('sequence')