def applyMethod(self ):
"""index the graph.
"""
self.info( "indexing of %s started" % self.mFilenameInputGraph )
self.info( "loading map_id2nid from %s" % self.mConfig.get( "files", "output_nids", "adda.nids" ))
infile = open( self.mConfig.get( "files", "output_nids", "adda.nids" ) )
map_id2nid = AddaIO.readMapId2Nid( infile,
storage = self.mConfig.get( "files", "storage_nids", "memory" ) )
infile.close()
infile = AddaIO.openStream( self.mFilenameInputGraph )
cadda.indexGraph( cadda.PairsDBNeighboursIterator(
self.mGraphIterator( infile, map_id2nid, self.mLogger ), self.mLogger ),
len(map_id2nid),
self.mFilenameOutputGraph,
self.mFilenameOutputIndex,
self.mLogger )
del map_id2nid
def applyMethod(self):
self.mInput = 0
self.mOutput = 0
self.mRemoved = 0
self.mDuplicates = 0
# use existing fasta file
iterator = FastaIterator(AddaIO.openStream(self.mFilenameInputFasta))
fasta = IndexedFasta.IndexedFasta(self.mFilenameOutputFasta, "w")
outfile = self.openOutputStream(self.mFilenameNids)
outfile.write("nid\tpid\thid\tlength\tsequence\n")
nid = 1
hids = set()
for seq in iterator:
self.mInput += 1
if len(seq.sequence) > self.mMaxSequenceLength:
self.mRemoved += 1
continue
hid = self.getHID(seq.sequence)
if hid in hids:
self.mDuplicates += 1
continue
hids.add(hid)
outfile.write("%s\t%s\t%s\t%i\t%s\n" %
(nid, seq.pid, hid, len(seq.sequence), seq.sequence))
fasta.addSequence(nid, seq.sequence)
nid += 1
self.mOutput += 1
fasta.close()
outfile.close()
def applyMethod(self):
"""index the graph.
"""
self.info("indexing of %s started" % self.mFilenameInputGraph)
self.info("loading map_id2nid from %s" %
self.mConfig.get("files", "output_nids", "adda.nids"))
infile = open(self.mConfig.get("files", "output_nids", "adda.nids"))
map_id2nid = AddaIO.readMapId2Nid(infile,
storage=self.mConfig.get(
"files", "storage_nids",
"memory"))
infile.close()
infile = AddaIO.openStream(self.mFilenameInputGraph)
cadda.indexGraph(
cadda.PairsDBNeighboursIterator(
self.mGraphIterator(infile, map_id2nid, self.mLogger),
self.mLogger), len(map_id2nid), self.mFilenameOutputGraph,
self.mFilenameOutputIndex, self.mLogger)
del map_id2nid
def applyMethod(self ):
self.mInput = 0
self.mOutput = 0
self.mRemoved = 0
self.mDuplicates = 0
# use existing fasta file
iterator = FastaIterator( AddaIO.openStream( self.mFilenameInputFasta) )
fasta = IndexedFasta.IndexedFasta( self.mFilenameOutputFasta, "w" )
outfile = self.openOutputStream(self.mFilenameNids)
outfile.write( "nid\tpid\thid\tlength\tsequence\n" )
nid = 1
hids = set()
for seq in iterator:
self.mInput += 1
if len( seq.sequence ) > self.mMaxSequenceLength:
self.mRemoved += 1
continue
hid = self.getHID( seq.sequence )
if hid in hids:
self.mDuplicates += 1
continue
hids.add(hid)
outfile.write( "%s\t%s\t%s\t%i\t%s\n" % (nid, seq.pid, hid, len(seq.sequence), seq.sequence) )
fasta.addSequence( nid, seq.sequence )
nid += 1
self.mOutput += 1
fasta.close()
outfile.close()
