def multi_ref(self):
refseq = d.refseq + d.chr22_chunk
u.create_seq_file(self.refseq_fn, refseq, qtype="fasta")
seq_len, span = 75, 500
f = open(self.refseq_fn)
raw_seqs = [r.seq.tostring() for r in Bio.SeqIO.parse(f, 'fasta')]
f.close()
raw_refseq = "".join(raw_seqs)
starts = [
0, # start of full refseq
len(raw_seqs[0]) - seq_len / 2, # bridges the two seqs
len(raw_refseq) - span # end of full refseq
]
pairs = u.generate_localized_pairs(raw_refseq,
starts,
seq_len,
span=span)
for i, fn in enumerate([self.read_fn, self.mate_fn]):
fp = open(fn, 'w')
Bio.SeqIO.write([x[i] for x in pairs], fp, 'fastq-illumina')
fp.close()
fp = open("%s.sanger" % fn, 'w')
Bio.SeqIO.write([x[i] for x in pairs], fp, 'fastq-sanger')
fp.close()
self.__do_pair_align_and_comparison(fastq_subfmt="fastq-illumina")
def auto_generated(self):
u.create_seq_file(self.refseq_fn, d.refseq, qtype="fasta")
n_seq, seq_len, span = 1000, 75, 500
refseq = Bio.SeqIO.parse(open(self.refseq_fn), 'fasta').next()
pairs = u.generate_pairs(refseq.seq, n_seq, seq_len, span=span)
for i, fn in enumerate([self.read_fn, self.mate_fn]):
fp = open(fn, 'w')
Bio.SeqIO.write([x[i] for x in pairs], fp, 'fastq-illumina')
fp.close()
fp = open("%s.sanger" % fn, 'w')
Bio.SeqIO.write([x[i] for x in pairs], fp, 'fastq-sanger')
fp.close()
self.__do_pair_align_and_comparison(fastq_subfmt="fastq-illumina")
def auto_generated(self):
u.create_seq_file(self.refseq_fn, d.refseq, qtype="fasta")
n_seq, seq_len, span = 1000, 75, 500
refseq = Bio.SeqIO.parse(open(self.refseq_fn), 'fasta').next()
pairs = u.generate_pairs(refseq.seq, n_seq, seq_len, span=span)
for i, fn in enumerate([self.read_fn, self.mate_fn]):
fp = open(fn, 'w')
Bio.SeqIO.write([x[i] for x in pairs], fp, 'fastq-illumina')
fp.close()
fp = open("%s.sanger" % fn, 'w')
Bio.SeqIO.write([x[i] for x in pairs], fp, 'fastq-sanger')
fp.close()
self.__do_pair_align_and_comparison(fastq_subfmt="fastq-illumina")
def multi_ref(self):
refseq = d.refseq + d.chr22_chunk
u.create_seq_file(self.refseq_fn, refseq, qtype="fasta")
seq_len, span = 75, 500
f = open(self.refseq_fn)
raw_seqs = [r.seq.tostring() for r in Bio.SeqIO.parse(f, 'fasta')]
f.close()
raw_refseq = "".join(raw_seqs)
starts = [
0, # start of full refseq
len(raw_seqs[0]) - seq_len/2, # bridges the two seqs
len(raw_refseq) - span # end of full refseq
]
pairs = u.generate_localized_pairs(raw_refseq, starts, seq_len, span=span)
for i, fn in enumerate([self.read_fn, self.mate_fn]):
fp = open(fn, 'w')
Bio.SeqIO.write([x[i] for x in pairs], fp, 'fastq-illumina')
fp.close()
fp = open("%s.sanger" % fn, 'w')
Bio.SeqIO.write([x[i] for x in pairs], fp, 'fastq-sanger')
fp.close()
self.__do_pair_align_and_comparison(fastq_subfmt="fastq-illumina")
def hand_picked_real_2(self):
u.create_seq_file(self.refseq_fn, d.mm_chr17_chunk, qtype="fasta")
for t, f in (("read", self.read_fn), ("mate", self.mate_fn)):
seq = getattr(d, "real_case_2_%s" % t)
u.create_seq_file(f, seq, qtype="fastq-sanger")
self.__do_pair_align_and_comparison(fastq_subfmt="fastq-sanger")
def hand_picked(self): u.create_seq_file(self.refseq_fn, d.refseq, qtype="fasta") u.create_seq_file(self.read_fn, d.hit_1_seq, qtype="fastq-sanger") u.create_seq_file(self.mate_fn, d.hit_2_seq, qtype="fastq-sanger") self.__do_pair_align_and_comparison()
def hand_picked_real_2(self):
u.create_seq_file(self.refseq_fn, d.mm_chr17_chunk, qtype="fasta")
for t, f in (("read", self.read_fn), ("mate", self.mate_fn)):
seq = getattr(d, "real_case_2_%s" % t)
u.create_seq_file(f, seq, qtype="fastq-sanger")
self.__do_pair_align_and_comparison(fastq_subfmt="fastq-sanger")
def hand_picked(self):
u.create_seq_file(self.refseq_fn, d.refseq, qtype="fasta")
u.create_seq_file(self.read_fn, d.hit_1_seq, qtype="fastq-sanger")
u.create_seq_file(self.mate_fn, d.hit_2_seq, qtype="fastq-sanger")
self.__do_pair_align_and_comparison()