def __init__(self, file):
self.__gtfreader = GTFReader(file)
from gtfIO import GTFReader, GTFWriter
from optparse import OptionParser
parser = OptionParser()
parser.add_option("--input",
dest="input",
help="A gtf file containing the RepeatMasked gtf annotation")
parser.add_option("--minlen", dest="minleng", help="minimum length")
parser.add_option("--output", dest="output", help="A gtf output file")
parser.add_option("--maxdiv",
dest="maxdiv",
help="minimum length",
default=99999999999999999)
(options, args) = parser.parse_args()
minleng = int(options.minleng)
w = GTFWriter(options.output)
for e in GTFReader(options.input):
leng = (e.end - e.start) + 1
if (leng >= minleng):
if (e.score < float(options.maxdiv)):
w.write(e)
w.close()
parser = OptionParser()
parser.add_option("--reference",dest="reference",help="A gtf file containing the reference annotation")
parser.add_option("--totest",dest="totest",help="A gtf file containing the annotation that should be tested")
(options, args) = parser.parse_args()
regionfilter={
"X": [0,22422827],
"2L": [0,22420241],
"2R": [387345,21146708],
"3L": [0,23825333],
"3R": [0,27905053],
"4" : [0,1350078]}
# chromosomes with incomplete reference annotation: 2L, 2R, 3L
# fine: 3R, X, 4
rawref=filterregion(GTFReader.readall(options.reference),regionfilter)
rawtotest=filterregion(GTFReader.readall(options.totest),regionfilter)
chref=convert_chrhash(rawref)
chtotest=convert_chrhash(rawtotest)
# ONLY TEST the chromosomes in the given list
tcboverlap,tcbref,tcbtotest=(0,0,0) # total count bases =tcb
tceoverlap,tceoverlap2,tceref,tcetotest=(0,0,0,0) # total count element =tcb
countRef=0
countTotest=0
countBRef=0
countBTotest=0
for chr in regionfilter.keys():
cref=chref[chr]
ctotest=chtotest[chr]
#intoverlapref=identifyOverlap(cref)
"--totest",
dest="totest",
help="A gtf file containing the annotation that should be tested")
(options, args) = parser.parse_args()
regionfilter = {
"X": [0, 22422827],
"2L": [0, 22420241],
"2R": [387345, 21146708],
"3L": [0, 23825333],
"3R": [0, 27905053],
"4": [0, 1350078]
}
# chromosomes with incomplete reference annotation: 2L, 2R, 3L
# fine: 3R, X, 4
rawref = filterregion(GTFReader.readall(options.reference), regionfilter)
rawtotest = filterregion(GTFReader.readall(options.totest), regionfilter)
chref = convert_chrhash(rawref)
chtotest = convert_chrhash(rawtotest)
# ONLY TEST the chromosomes in the given list
tcboverlap, tcbref, tcbtotest = (0, 0, 0) # total count bases =tcb
tceoverlap, tceoverlap2, tceref, tcetotest = (0, 0, 0, 0
) # total count element =tcb
countRef = 0
countTotest = 0
countBRef = 0
countBTotest = 0
for chr in regionfilter.keys():
cref = chref[chr]
ctotest = chtotest[chr]
help="A gtf file containing the reference annotation")
parser.add_option(
"--totest",
dest="totest",
help="A gtf file containing the annotation that should be tested")
parser.add_option(
"--output",
dest="output",
help=
"The output file containing novel TE insertions not in the reference genome"
)
(options, args) = parser.parse_args()
# chromosomes with incomplete reference annotation: 2L, 2R, 3L
# fine: 3R, X, 4
rawref = GTFReader.readall(options.reference)
rawtotest = GTFReader.readall(options.totest)
chref = convert_chrhash(rawref)
chtotest = convert_chrhash(rawtotest)
ofh = GTFWriter(options.output)
for chr in ["X", "2L", "2R", "3L", "3R", "4"]:
cref = chref[chr]
ctotest = chtotest[chr]
ne = novelentries(cref, ctotest)
for n in ne:
ofh.write(n)
ofh.close()
e.target=target
return e
raise StopIteration
parser = OptionParser()
parser.add_option("--reference",dest="reference",help="A gtf file containing the reference annotation")
parser.add_option("--totest",dest="totest",help="A gtf file containing the annotation that should be tested")
parser.add_option("--output",dest="output",help="The output file containing novel TE insertions not in the reference genome")
(options, args) = parser.parse_args()
# chromosomes with incomplete reference annotation: 2L, 2R, 3L
# fine: 3R, X, 4
rawref=GTFReader.readall(options.reference)
rawtotest=GTFReader.readall(options.totest)
chref=convert_chrhash(rawref)
chtotest=convert_chrhash(rawtotest)
ofh=GTFWriter(options.output)
for chr in ["X","2L","2R","3L","3R","4"]:
cref=chref[chr]
ctotest=chtotest[chr]
ne=novelentries(cref,ctotest)
for n in ne:
ofh.write(n)
ofh.close()