def __init__( self, outFolder='.', clusterLimit=20, verbose=1,
log=None, silent=0 ):
"""
@param outFolder: project folder (results are put into subfolder) ['.']
@type outFolder: str
@param clusterLimit: maximal number of returned sequence clusters
(default: 20)
@type clusterLimit: int
@param verbose: keep temporary files (default: 1)
@type verbose: 1|0
@param log: log file instance, if None, STDOUT is used (default: None)
@type log: LogFile
@param silent: don't print messages to STDOUT is used (default: 0)
@type silent: 1|0
"""
SequenceSearcher.__init__( self, outFolder=outFolder, verbose=1 )
self.ex_pdb = re.compile( 'pdb\|([A-Z0-9]{4})\|([A-Z]*)' )
self.ex_gb = re.compile( 'gi\|([0-9]+)\|')
self.ex_resolution = re.compile(\
'REMARK 2 RESOLUTION\. *([0-9\.]+|NOT APPLICABLE)' )
self.prepareFolders()
self.verbose = verbose
self.silent = silent
self.log = log or StdLog()
## the maximal number of clusters to return
self.clusterLimit = clusterLimit
def prepareFolders(self):
"""
Create folders needed by this class.
"""
SequenceSearcher.prepareFolders(self)
if not os.path.exists(self.outFolder + self.F_ALL):
os.mkdir(self.outFolder + self.F_ALL)
if not os.path.exists(self.outFolder + self.F_NR):
os.mkdir(self.outFolder + self.F_NR)
def prepareFolders( self ):
"""
Create folders needed by this class.
"""
SequenceSearcher.prepareFolders( self )
if not os.path.exists( self.outFolder + self.F_ALL ):
os.mkdir( self.outFolder + self.F_ALL )
if not os.path.exists( self.outFolder + self.F_NR ):
os.mkdir( self.outFolder + self.F_NR )
def fastaRecordFromId(self, db, id, chain=''):
"""
Use::
fastaRecordFromId( db, id ) -> Bio.Fasta.Record
@param db: database
@type db: str
@param id: sequence database ID
@type id: str
@return: fasta record
@rtype: Bio.SeqRecord.SeqRecord
@raise BlastError: if can't fetch fasta record from database
"""
if chain == '':
return SequenceSearcher.fastaRecordFromId(self, db, id)
cmd = settings.fastacmd_bin + " -d %s -s 'pdb|%s|%s'" \
% (db, id, chain)
err, o = commands.getstatusoutput(cmd)
if err:
EHandler.warning('%s returned error: %r' % (cmd, err))
raise BlastError(err)
try:
frecord = SeqIO.parse(cStringIO.StringIO(o), 'fasta').next()
frecord.id = str(id)
except StopIteration:
raise InternalError, \
"Couldn't fetch fasta record %s from database %s" % (id,db)
return frecord
def fastaRecordFromId( self, db, id, chain='' ):
"""
Use::
fastaRecordFromId( db, id ) -> Bio.Fasta.Record
@param db: database
@type db: str
@param id: sequence database ID
@type id: str
@return: fasta record
@rtype: Bio.SeqRecord.SeqRecord
@raise BlastError: if can't fetch fasta record from database
"""
if chain == '':
return SequenceSearcher.fastaRecordFromId( self, db, id )
cmd = settings.fastacmd_bin + " -d %s -s 'pdb|%s|%s'" \
% (db, id, chain)
err, o = commands.getstatusoutput( cmd )
if err:
EHandler.warning('%s returned error: %r' % (cmd, err) )
raise BlastError( err )
try:
frecord = SeqIO.parse( cStringIO.StringIO(o), 'fasta').next()
frecord.id = str(id)
except StopIteration:
raise InternalError, \
"Couldn't fetch fasta record %s from database %s" % (id,db)
return frecord