This script converts a fasta file and a given fasta reference file to
a file in VCF. A reference genome can be provided in fasta format as
input. If no reference is given, the first sequence in the fasta file
will be used as reference.
"""
import argparse
import libPoMo.fasta as fa # noqa
import libPoMo.vcf as vcf # noqa
ver = "1.0"
parser = argparse.ArgumentParser(prog="FastaToVCF.py",
description="convert fasta to VCF;" +
" script version" + ver)
parser.add_argument("fastafile", help="path to fasta file")
parser.add_argument("output", help="name of VCF output file")
parser.add_argument("-r",
"--reference",
help="path to reference genome in fasta format")
args = parser.parse_args()
faSeq = fa.open_seq(args.fastafile)
if args.reference is not None:
faRef = fa.open_seq(args.reference)
refSeq = faRef.get_seq_by_id(0)
else:
refSeq = faSeq.get_seq_by_id(0)
fa.save_as_vcf(faSeq, refSeq, args.output)
a file in VCF. A reference genome can be provided in fasta format as
input. If no reference is given, the first sequence in the fasta file
will be used as reference.
"""
import argparse
import libPoMo.fasta as fa # noqa
import libPoMo.vcf as vcf # noqa
ver = "1.0"
parser = argparse.ArgumentParser(prog="FastaToVCF.py",
description="convert fasta to VCF;" +
" script version" + ver)
parser.add_argument("fastafile",
help="path to fasta file")
parser.add_argument("output",
help="name of VCF output file")
parser.add_argument("-r", "--reference",
help="path to reference genome in fasta format")
args = parser.parse_args()
faSeq = fa.open_seq(args.fastafile)
if args.reference is not None:
faRef = fa.open_seq(args.reference)
refSeq = faRef.get_seq_by_id(0)
else:
refSeq = faSeq.get_seq_by_id(0)
fa.save_as_vcf(faSeq, refSeq, args.output)
print("Read in test sequence ", test_sequence, '.', sep='')
seq = fa.open_seq(test_sequence)
print("Print sequence information.")
seq.print_info()
######################################################################
print("\n##################################################")
print("Test FaStream object.")
faStr = fa.init_seq(test_sequence)
faStr.print_info(maxB=None)
while faStr.read_next_seq() is not None:
faStr.print_info()
faStr.close()
######################################################################
print("\n##################################################")
test_sequence = "data/fasta-sample-wolfs.dat"
ref_sequence = "data/fasta-reference-wolf.dat"
fn = "vcf-test-tmp.dat"
print("Compare ", test_sequence, " to ", ref_sequence, '.', sep='')
faSeq = fa.open_seq(test_sequence)
faRef = fa.open_seq(ref_sequence)
refSeq = faRef.get_seq_by_id(0)
fa.save_as_vcf(faSeq, refSeq, fn)
with open(fn) as file:
for line in file:
print(line, end='')
os.remove(fn)
print("\n##################################################")
print("Read in test sequence ", test_sequence, '.', sep='')
seq = fa.open_seq(test_sequence)
print("Print sequence information.")
seq.print_info()
######################################################################
print("\n##################################################")
print("Test FaStream object.")
faStr = fa.init_seq(test_sequence)
faStr.print_info(maxB=None)
while faStr.read_next_seq() is not None:
faStr.print_info()
faStr.close()
######################################################################
print("\n##################################################")
test_sequence = "data/fasta-sample-wolfs.dat"
ref_sequence = "data/fasta-reference-wolf.dat"
fn = "vcf-test-tmp.dat"
print("Compare ", test_sequence, " to ", ref_sequence, '.', sep='')
faSeq = fa.open_seq(test_sequence)
faRef = fa.open_seq(ref_sequence)
refSeq = faRef.get_seq_by_id(0)
fa.save_as_vcf(faSeq, refSeq, fn)
with open(fn) as file:
for line in file:
print(line, end='')
os.remove(fn)
