def merge_fastq(fastq_path1, fastq_path2):
combine_str = ""
if isCompressed(fastq_path1):
with gzip.open(fastq_path1, "rt") as handle1:
for read in SeqIO.parse(handle1, "fastq"):
combine_str = combine_str + str(read.format("fastq"))
if isCompressed(fastq_path2):
with gzip.open(fastq_path1, "rt") as handle2:
for read in SeqIO.parse(handle2, "fastq"):
combine_str = combine_str + str(read.format("fastq"))
else:
with open(fastq_path2, "rt") as handle2:
for read in SeqIO.parse(handle2, "fastq"):
combine_str = combine_str + str(read.format("fastq"))
else:
with open(fastq_path1, "rt") as handle1:
for read in SeqIO.parse(handle1, "fastq"):
combine_str = combine_str + str(read.format("fastq"))
if isCompressed(fastq_path2):
with gzip.open(fastq_path2, "rt") as handle2:
for read in SeqIO.parse(handle2, "fastq"):
combine_str = combine_str + str(read.format("fastq"))
else:
with open(fastq_path2, "rt") as handle2:
for read in SeqIO.parse(handle2, "fastq"):
combine_str = combine_str + str(read.format("fastq"))
return combine_str
def trim_quality(filename, limit):
r_str = ""
if isCompressed(filename):
with gzip.open(filename, "rt") as handle:
for read in SeqIO.parse(handle, "fastq"):
if min(read.letter_annotations["phred_quality"]) >= limit:
r_str = r_str + str(read.id) + "\n"
r_str = r_str + str(read.seq) + "\n"
r_str = r_str + str(
read.format("fastq").split("\n")[2]) + "\n"
r_str = r_str + str(
read.format("fastq").split("\n")[3]) + "\n"
else:
with open(filename, "rt") as handle:
for read in SeqIO.parse(handle, "fastq"):
if min(read.letter_annotations["phred_quality"]) >= limit:
r_str = r_str + str(read.id) + "\n"
r_str = r_str + str(read.seq) + "\n"
r_str = r_str + str(
read.format("fastq").split("\n")[2]) + "\n"
r_str = r_str + str(
read.format("fastq").split("\n")[3]) + "\n"
return r_str
def fetch_all_reads(filename, option):
filter_str = ""
if isCompressed(filename):
with gzip.open(filename, "rt") as handle:
#filterstr=""
if option == 0:
for read in SeqIO.parse(handle, "fastq"):
filter_str = filter_str + str(read.format("fastq"))
break
elif option == 1:
for read in SeqIO.parse(handle, "fastq"):
filter_str = filter_str + str(read.seq) + "\n"
elif option == 2:
for read in SeqIO.parse(handle, "fastq"):
filter_str = filter_str + str(read.id) + "\n"
elif option == 3:
for read in SeqIO.parse(handle, "fastq"):
filter_str = filter_str + str(
read.format("fastq").split("\n")[3]) + "\n"
else:
for read in SeqIO.parse(handle, "fastq"):
filter_str = filter_str + str(read.id) + "\n"
filter_str = filter_str + str(read.seq) + "\n"
else:
with open(filename, "rt") as handle:
#filterstr=""
if option == 0:
for read in SeqIO.parse(handle, "fastq"):
filter_str = filter_str + str(read.format("fastq"))
print("im in option 0")
#print(filter_str)
elif option == 1:
for read in SeqIO.parse(handle, "fastq"):
print("im in option 1")
filter_str = filter_str + str(read.seq) + "\n"
elif option == 2:
for read in SeqIO.parse(handle, "fastq"):
filter_str = filter_str + str(read.id) + "\n"
elif option == 3:
for read in SeqIO.parse(handle, "fastq"):
filter_str = filter_str + str(
read.format("fastq").split("\n")[3]) + "\n"
else:
for read in SeqIO.parse(handle, "fastq"):
filter_str = filter_str + str(read.id) + "\n"
filter_str = filter_str + str(read.seq) + "\n"
return filter_str
def trimoffprimer(filename, primersequence):
trim_str = ""
if isCompressed(filename):
with gzip.open(filename, "rt") as handle:
for read in SeqIO.parse(handle, "fastq"):
trim_str = trim_str + str(trim_primer(read, primersequence))
#return trim_str
else:
with open(filename, "rt") as handle:
for read in SeqIO.parse(handle, "fastq"):
trim_str = trim_str + str(trim_primer(read, primersequence))
return trim_str
def isFastq(filename):
#status=True
if isCompressed(filename):
try:
with gzip.open(filename, "rt") as handle:
read = SeqIO.parse(handle, "fastq")
return any(read)
except:
return False
else:
try:
with open(filename, "rt") as handle:
read = SeqIO.parse(handle, "fastq")
return any(read)
except:
return False
def search(filename, sequence):
filesearch = ""
if isCompressed(filename):
with gzip.open(filename, "rt") as handle:
for read in SeqIO.parse(handle, "fastq"):
if (sequence in read.seq):
filesearch = str(filesearch + read.id + "\n" + read.seq +
"\n")
return filesearch
else:
with open(filename, "rt") as handle:
for read in SeqIO.parse(handle, "fastq"):
if sequence in read.seq:
filesearch = str(filesearch + read.id + "\n" + read.seq +
"\n")
return filesearch
#a=search("t1.fastq.gz","CCGC")
#print(a)
#print(type(a))
#file_save(a)
def statistics(filename):
count=1
rlength=0
gc=0
if isCompressed(filename):
with gzip.open(filename, "rt") as handle:
for read in SeqIO.parse(handle,"fastq"):
count=count+1
gc=gc+GC(read.seq)
rlength=rlength+len(read.seq)
count=count-1
average=round(rlength/count)
gcper=round(gc/count)
return [count,average,"Compressed",gcper]
else:
with open(filename, "rt") as handle:
for read in SeqIO.parse(handle,"fastq"):
count=count+1
gc=gc+GC(read.seq)
rlength=rlength+len(read.seq)
count=count-1
average=round(rlength/count)
gcper=round(gc/count)
return [count,average,"Simple Text",gcper]
def fetch_head_reads_only(filename, option):
filter_str = ""
count = 0
#record_length = len(list(SeqIO.parse("test.fastq", "fastq")))
if isCompressed(filename):
with gzip.open(filename, "rt") as handle:
#filterstr=""
if option == 0:
for read in SeqIO.parse(handle, "fastq"):
filter_str = filter_str + str(read.format("fastq"))
count = count + 1
if count >= 10:
break
elif option == 1:
for read in SeqIO.parse(handle, "fastq"):
filter_str = filter_str + str(read.seq) + "\n"
if count >= 10:
break
elif option == 2:
for read in SeqIO.parse(handle, "fastq"):
filter_str = filter_str + str(read.id) + "\n"
if count >= 10:
break
elif option == 3:
for read in SeqIO.parse(handle, "fastq"):
filter_str = filter_str + str(
read.format("fastq").split("\n")[3]) + "\n"
if count >= 10:
break
else:
for read in SeqIO.parse(handle, "fastq"):
filter_str = filter_str + str(read.id) + "\n"
filter_str = filter_str + str(read.seq) + "\n"
if count >= 10:
break
else:
with open(filename, "rt") as handle:
#filterstr=""
if option == 0:
for read in SeqIO.parse(handle, "fastq"):
filter_str = filter_str + str(read.format("fastq"))
count = count + 1
print(count)
if count >= 10:
break
elif option == 1:
for read in SeqIO.parse(handle, "fastq"):
filter_str = filter_str + str(read.seq) + "\n"
if count >= 10:
break
elif option == 2:
for read in SeqIO.parse(handle, "fastq"):
filter_str = filter_str + str(read.id) + "\n"
if count >= 10:
break
elif option == 3:
for read in SeqIO.parse(handle, "fastq"):
filter_str = filter_str + str(
read.format("fastq").split("\n")[3]) + "\n"
if count >= 10:
break
else:
for read in SeqIO.parse(handle, "fastq"):
filter_str = filter_str + str(read.id) + "\n"
filter_str = filter_str + str(read.seq) + "\n"
if count >= 10:
break
return filter_str
def fetch_inrange_reads_only(filename, option, l_range, u_range):
filter_str = ""
count = 0
print("hello")
print(l_range, u_range)
record_length = len(list(SeqIO.parse("test.fastq", "fastq")))
if isCompressed(filename):
with gzip.open(filename, "rt") as handle:
#filterstr=""
if option == 0:
for read in SeqIO.parse(handle, "fastq"):
if count >= l_range - 1 and count <= u_range + 1:
filter_str = filter_str + str(read.format("fastq"))
count = count + 1
count = count + 1
elif option == 1:
for read in SeqIO.parse(handle, "fastq"):
if count >= l_range - 1 and count <= u_range + 1:
filter_str = filter_str + str(read.seq) + "\n"
count = count + 1
count = count + 1
elif option == 2:
for read in SeqIO.parse(handle, "fastq"):
if count >= l_range - 1 and count <= u_range + 1:
filter_str = filter_str + str(read.id) + "\n"
count = count + 1
count = count + 1
elif option == 3:
for read in SeqIO.parse(handle, "fastq"):
if count >= l_range - 1 and count <= u_range + 1:
filter_str = filter_str + str(
read.format("fastq").split("\n")[3]) + "\n"
count = count + 1
count = count + 1
else:
for read in SeqIO.parse(handle, "fastq"):
if count >= l_range - 1 and count <= u_range + 1:
filter_str = filter_str + str(read.id) + "\n"
filter_str = filter_str + str(read.seq) + "\n"
count = count + 1
count = count + 1
else:
with open(filename, "rt") as handle:
if option == 0:
for read in SeqIO.parse(handle, "fastq"):
if count >= l_range - 1 and count <= u_range + 1:
filter_str = filter_str + str(read.format("fastq"))
count = count + 1
count = count + 1
elif option == 1:
for read in SeqIO.parse(handle, "fastq"):
if count >= l_range - 1 and count <= u_range + 1:
filter_str = filter_str + str(read.seq) + "\n"
count = count + 1
count = count + 1
elif option == 2:
for read in SeqIO.parse(handle, "fastq"):
if count >= l_range - 1 and count <= u_range + 1:
filter_str = filter_str + str(read.id) + "\n"
count = count + 1
count = count + 1
elif option == 3:
for read in SeqIO.parse(handle, "fastq"):
if count >= l_range - 1 and count <= u_range + 1:
filter_str = filter_str + str(
read.format("fastq").split("\n")[3]) + "\n"
count = count + 1
count = count + 1
else:
for read in SeqIO.parse(handle, "fastq"):
if count >= l_range - 1 and count <= u_range + 1:
filter_str = filter_str + str(read.id) + "\n"
filter_str = filter_str + str(read.seq) + "\n"
count = count + 1
count = count + 1
return filter_str