#Downsample a library
import sys
from nucio import typeify, fileIterator
from seqio import iteratorFromExtension, recordToString, seqlen
if not len(sys.argv) == 5:
sys.exit("Usage: downsample.py genome_size desired_cov input.{fa,fq} output.{fa,fq}\n")
types = [int, float, str, str]
sysins = sys.argv[1:len(types)+1]
(genome_size, target_cov, infn, outfn) = typeify(sysins,types)
max_bases = genome_size * target_cov
total_bases = 0
with open(outfn, "w") as of:
for record in fileIterator(infn,iteratorFromExtension(infn)):
length = seqlen(record)
if "N" in record.seq:
continue
if total_bases > max_bases:
break
of.write(recordToString(record))
of.write("\n")
total_bases += length
from seqio import iteratorFromExtension, recordToString, FastqRecord
from nucio import fileIterator
from args import parseArgs, getHelpStr, CLArgument
description = (
"Usage: deplex_pb.py [options] file1.{fa,fq} [file2.{fa,fq} ..]\n\n"
"Deplexes a file based on some delimiter")
argument_list = [["delim", "delim", str, "/", "Delimiter to split the input"]]
arguments = map(CLArgument._make, argument_list)
if not len(sys.argv) >= 2:
sys.exit(getHelpStr(description, arguments) + "\n")
(p_arg_map, args_remaining) = parseArgs(sys.argv[1:], arguments)
its = map(iteratorFromExtension, args_remaining)
file_its = starmap(fileIterator, izip(args_remaining, its))
fh_h = {}
for entry in chain.from_iterable(file_its):
h = entry.name.split(p_arg_map["delim"])[0]
ext = ".fastq" if isinstance(entry, FastqRecord) else ".fasta"
if not h in fh_h:
fh_h[h] = open(h + ext, "w")
fh_h[h].write(recordToString(entry))
fh_h[h].write("\n")
#!/usr/bin/env python
import sys
from seqio import iteratorFromExtension, recordToString
from nucio import fileIterator
from misc import reverse_complement
if not len(sys.argv) == 2:
sys.exit("reverseComplement.py in.{fa,fq}")
f = sys.argv[1]
for record in fileIterator(f, iteratorFromExtension(f)):
print recordToString(record._replace(seq=reverse_complement(record.seq)))
#!/usr/bin/env python
import sys
from seqio import iteratorFromExtension, recordToString
from nucio import fileIterator
from misc import reverse_complement
if not len(sys.argv) == 2:
sys.exit("reverseComplement.py in.{fa,fq}")
f = sys.argv[1]
for record in fileIterator(f,iteratorFromExtension(f)):
print recordToString(record._replace(seq=reverse_complement(record.seq)))
from seqio import iteratorFromExtension, recordToString, FastqRecord
from nucio import fileIterator
from args import parseArgs, getHelpStr, CLArgument
description = (
"Usage: deplex_pb.py [options] file1.{fa,fq} [file2.{fa,fq} ..]\n\n" "Deplexes a file based on some delimiter"
)
argument_list = [["delim", "delim", str, "/", "Delimiter to split the input"]]
arguments = map(CLArgument._make, argument_list)
if not len(sys.argv) >= 2:
sys.exit(getHelpStr(description, arguments) + "\n")
(p_arg_map, args_remaining) = parseArgs(sys.argv[1:], arguments)
its = map(iteratorFromExtension, args_remaining)
file_its = starmap(fileIterator, izip(args_remaining, its))
fh_h = {}
for entry in chain.from_iterable(file_its):
h = entry.name.split(p_arg_map["delim"])[0]
ext = ".fastq" if isinstance(entry, FastqRecord) else ".fasta"
if not h in fh_h:
fh_h[h] = open(h + ext, "w")
fh_h[h].write(recordToString(entry))
fh_h[h].write("\n")