def add(self, header, sequence, force=False):
key = hashSequence(sequence)
if key in self.data:
logging.info(
f"blast::add:Bouncing {header}, sequence already stored")
return
self._buffer.append((header, sequence, key))
def _remove_from_mfasta(self):
"""Parse current multifasta and just keep sequence that are not in _delete_buffer in _buffer. The _buffer sequences can be then rewrite in a multifasta.
"""
to_delete_headers = [buf[0] for buf in self._delete_buffer]
for header, seq, _id in zFastaReader(self.fastaBufferFile):
key = hashSequence(seq)
_header = f">{header}"
if not _header in to_delete_headers:
self._buffer.append(
(_header, seq, key)
) # Add sequences that we want to keep to buffer for rewriting
def remove(self, header: str, sequence: str):
"""Remove sequence from blast database
:param header: Sequence header
:type header: str
:param sequence: Nucleotide sequence
:type sequence: str
"""
key = hashSequence(sequence)
if not key in self.data:
logging.warn(
f"blast::remove:Fasta sequence {header} doesn't exist in blast database"
)
else:
self._delete_buffer.append((header, key))
self._remove_from_mfasta()
def _index(self):
if self.registry['pkl']:
return self._restoreIndex(
f"{self.location}/{self.registry['pkl']}")
logging.info(f"Building index on {self.fastaFile}, "
f" this may take a while...")
data = {}
with zFile(self.fastaFile) as handle:
for genome_seqrecord in SeqIO.parse(handle, "fasta"):
genome_seq = genome_seqrecord.seq
header = genome_seqrecord.id
_id = hashSequence(str(genome_seq))
data[_id] = header
logging.info(f"{len(data.keys())} fasta records successfully indexed")
return data
def get(self, **kwargs):
if 'seq' in kwargs:
return self[hashSequence(seq)]