def dbORF( inFileName, nb=0, size=0, outFileName="" ):
inFile = open( inFileName )
if outFileName == "":
outFileName = inFileName + ".orf.map"
outFile = open( outFileName, "w" )
seq = Bioseq()
seq_out = Bioseq()
numseq = 0
while 1:
seq.read( inFile )
if seq.sequence == None:
break
seq.upCase()
numseq = numseq + 1
print 'sequence #',numseq,'=',seq.getLength(),'[',seq.header[0:40],'...]'
orf = seq.findORF()
best_orf = []
for i in orf.keys():
l = len(orf[i])
for j in xrange(1,l):
start = orf[i][j-1] + 4
end = orf[i][j] + 3
if end - start >= size:
best_orf.append( ( end-start, i+1, start, end ) )
seq.sequence = seq.complement()
orf = seq.findORF()
seqlen = seq.getLength()
for i in orf.keys():
l = len(orf[i])
for j in xrange(1,l):
start = seqlen - orf[i][j-1] - 3
end = seqlen - orf[i][j] - 2
if start - end >= size:
best_orf.append( ( start-end, (i+1)*-1, start, end ) )
best_orf.sort()
best_orf.reverse()
l = len(best_orf)
if nb > l or nb == 0 :
nb = l
for i in xrange(0,nb):
print best_orf[i]
outFile.write("%s\t%s\t%d\t%d\n"%("ORF|"+str(best_orf[i][1])+\
"|"+str(best_orf[i][0]),seq.header,
best_orf[i][2],best_orf[i][3]))
inFile.close()
outFile.close()
return 0
def dbTraduit(inFileName,phase=0,complement='T',pep_filename=""): """ deprecated """ file_db=open(inFileName) if pep_filename=="": pep_filename=inFileName+'.pep' file_pep=open(pep_filename,'w') seq=Bioseq() seq_out=Bioseq() numseq=0 while 1: seq.read(file_db) if seq.sequence==None: break numseq=numseq+1 print 'sequence #',numseq,'=',seq.getLength(),\ '[',seq.header[0:40],'...]' if phase>=0 : if phase==1 or phase==0 : seq_out.sequence=seq.traduit(1) seq_out.header=seq.header+" (phase 1)" seq_out.write(file_pep) if phase==2 or phase==0 : seq_out.sequence=seq.traduit(2) seq_out.header=seq.header+" (phase 2)" seq_out.write(file_pep) if phase==3 or phase==0 : seq_out.sequence=seq.traduit(3) seq_out.header=seq.header+" (phase 3)" seq_out.write(file_pep) if complement=='T' or phase<0 : seq.sequence=seq.complement() if phase==-1 or phase==0 : seq_out.sequence=seq.traduit(1) seq_out.header=seq.header+" (phase -1)" seq_out.write(file_pep) if phase==-2 or phase==0 : seq_out.sequence=seq.traduit(2) seq_out.header=seq.header+" (phase -2)" seq_out.write(file_pep) if phase==-3 or phase==0 : seq_out.sequence=seq.traduit(3) seq_out.header=seq.header+" (phase -3)" seq_out.write(file_pep) file_db.close() file_pep.close()
def sortSequencesByIncreasingLength( inFile, outFile, verbose=0 ):
"""
Save sequences in 'inFile' into 'outFile' sorted by their length in increasing order.
"""
if verbose > 0:
print "sort sequences by increasing length"
sys.stdout.flush()
if not os.path.exists( inFile ):
print "ERROR: file '%s' doesn't exist" % ( inFile )
sys.exit(1)
# read each seq one by one
# save them in distinct temporary files
# with their length in the name
inFileHandler = open( inFile, "r" )
bs = Bioseq()
countSeq = 0
while True:
bs.read( inFileHandler )
if bs.header == None:
break
countSeq += 1
tmpFile = "%ibp_%inb" % ( bs.getLength(), countSeq )
bs.save( tmpFile )
if verbose > 1:
print "%s (%i bp) saved in '%s'" % ( bs.header, bs.getLength(), tmpFile )
bs.header = ""
bs.sequence = ""
inFileHandler.close()
# sort temporary file names
# concatenate them into the output file
if os.path.exists( outFile ):
os.remove( outFile )
lFiles = glob.glob( "*bp_*nb" )
lFiles.sort( key=lambda s:int(s.split("bp_")[0]) )
for fileName in lFiles:
cmd = "cat %s >> %s" % ( fileName, outFile )
returnValue = os.system( cmd )
if returnValue != 0:
print "ERROR while concatenating '%s' with '%s'" % ( fileName, outFile )
sys.exit(1)
os.remove( fileName )
return 0
def dbComplement(inFileName,comp_filename=""): """ deprecated """ file_db=open(inFileName) if comp_filename=="": comp_filename=inFileName+'.comp' file_comp=open(comp_filename,'w') seq=Bioseq() numseq=0 while 1: seq.read(file_db) if seq.sequence==None: break numseq=numseq+1 print 'sequence #',numseq,'=',seq.getLength(),'[',seq.header[0:40],'...]' seq.sequence=seq.complement() seq.header=seq.header+" (complement!)" seq.write(file_comp) file_db.close() file_comp.close()