def writeout(clusters, output_filename):
w = BowTieWriter(output_filename)
for cid, c in clusters.iteritems():
qual = "".join(chr(int(round(x+33))) for x in np.array(c['qual']) / np.array(c['abun']))
r = {'ID': cid, 'seq': c['seq'], 'qual': qual, 'strand': '+', \
'ref': str(c['cycle']), 'offset': max(c['abun'])}
w.write(r)
w.close()
def main(fq1, fq2, output_prefix, abundance_filename):
abundance = {}
if abundance_filename is None:
abundance = defaultdict(lambda: 1)
else:
with open(abundance_filename) as f:
for line in f:
_id, _count = line.strip().split('\t')
abundance[_id] = int(_count)
matchf = BowTieWriter(output_prefix + '.overlap.aligned')
unf1 = FastqWriter(output_prefix + '.overlap.1.unaligned')
unf2 = FastqWriter(output_prefix + '.overlap.2.unaligned')
total = 0
total_expanded = 0
aligned = 0
aligned_expanded = 0
for r1, r2 in FastqReaderPaired(fq1, fq2):
realid = r1['ID'][:r1['ID'].find('/')]
total += 1
total_expanded += abundance[realid]
if find_overlap(r1, r2, matchf, unf1, unf2): #overlap found
aligned += 1
aligned_expanded += abundance[realid]
with open(output_prefix + '.overlap.log', 'w') as f:
p = aligned * 100. / total
f.write("# reads processed: {0}\n".format(total))
f.write(
"# reads with at least one reported alignment: {0} ({1:.2f}%)\n".
format(aligned, p))
f.write("# reads that failed to align: {0} ({1:.2f}%)\n".format(
total - aligned, 100 - p))
f.write("Reported {0} paired-end alignments to 1 output stream(s)\n".
format(aligned))
with open(output_prefix + '.overlap.log_expanded', 'w') as f:
p = aligned_expanded * 100. / total_expanded
f.write("# reads processed: {0}\n".format(total_expanded))
f.write(
"# reads with at least one reported alignment: {0} ({1:.2f}%)\n".
format(aligned_expanded, p))
f.write("# reads that failed to align: {0} ({1:.2f}%)\n".format(
total_expanded - aligned_expanded, 100 - p))
f.write("Reported {0} paired-end alignments to 1 output stream(s)\n".
format(aligned_expanded))
matchf.close()
unf1.close()
unf2.close()
def main(fq1, fq2, output_prefix, abundance_filename):
abundance = {}
if abundance_filename is None:
abundance = defaultdict(lambda: 1)
else:
with open(abundance_filename) as f:
for line in f:
_id, _count = line.strip().split('\t')
abundance[_id] = int(_count)
matchf = BowTieWriter(output_prefix + '.overlap.aligned')
unf1 = FastqWriter(output_prefix + '.overlap.1.unaligned')
unf2 = FastqWriter(output_prefix + '.overlap.2.unaligned')
total = 0
total_expanded = 0
aligned = 0
aligned_expanded = 0
for r1, r2 in FastqReaderPaired(fq1, fq2):
realid = r1['ID'][:r1['ID'].find('/')]
total += 1
total_expanded += abundance[realid]
if find_overlap(r1, r2, matchf, unf1, unf2): #overlap found
aligned += 1
aligned_expanded += abundance[realid]
with open(output_prefix + '.overlap.log', 'w') as f:
p = aligned*100./total
f.write("# reads processed: {0}\n".format(total))
f.write("# reads with at least one reported alignment: {0} ({1:.2f}%)\n".format(aligned,p))
f.write("# reads that failed to align: {0} ({1:.2f}%)\n".format(total-aligned,100-p))
f.write("Reported {0} paired-end alignments to 1 output stream(s)\n".format(aligned))
with open(output_prefix + '.overlap.log_expanded', 'w') as f:
p = aligned_expanded*100./total_expanded
f.write("# reads processed: {0}\n".format(total_expanded))
f.write("# reads with at least one reported alignment: {0} ({1:.2f}%)\n".format(aligned_expanded,p))
f.write("# reads that failed to align: {0} ({1:.2f}%)\n".format(total_expanded-aligned_expanded,100-p))
f.write("Reported {0} paired-end alignments to 1 output stream(s)\n".format(aligned_expanded))
matchf.close()
unf1.close()
unf2.close()
def filter_low_qual_seqs(gz_filename1, gz_filename2, phred_offset, phred_cutoff):
"""
Takes a BowTie-style gzipped file (ex: .aligned.composite.gz)
and retain only seqs that have every base phred >= <cutoff>
Outputs: .phred<cutoff>_passed for both files
"""
assert phred_offset >= 0
assert phred_cutoff >= 0
bad = 0
good = 0
start_t = time.time()
print gz_filename1, gz_filename2
f1 = BowTieWriter(gz_filename1 + ".phred{0}_passed".format(phred_cutoff), 'w')
f2 = BowTieWriter(gz_filename2 + ".phred{0}_passed".format(phred_cutoff), 'w')
for r1, r2 in itertools.izip(BowTieReader(gz_filename1,False), BowTieReader(gz_filename2,False)):
if all(ord(x)-phred_offset >= phred_cutoff for x in r1['qual']) and \
all(ord(x)-phred_offset >= phred_cutoff for x in r2['qual']):
good += 1
f1.write(r1)
f2.write(r2)
else:
bad += 1
with open(gz_filename1 + ".phred{0}_passed.log".format(phred_cutoff), 'w') as f:
f.write("Running filter_low_qual_seq took {0} secs\n".format(time.time()-start_t))
f.write("Input: " + gz_filename1 + ',' + gz_filename2 + '\n')
f.write("PhredCutoff: " + str(phred_cutoff) + '\n')
f.write("RemovedDueToLowQual: " + str(bad) + '\n')
f.write("RemainingTotal: " + str(good) + '\n')
def split_fasta_by_otu(fasta_filename, bowtie_gz_filename, otu_filename,
output_dir):
"""
For each OTU, create a subdir <output_dir>/<cluster_index> and
put it in the OTU's fasta and bowtie (gzipped)
"""
otu = {}
fa_d = {}
bw_d = {}
cids = set()
with open(otu_filename) as f:
for line in f:
raw = line.strip().split()
cid = raw[0]
cids.add(cid)
if not os.path.exists(os.path.join(output_dir, cid)):
os.mkdir(os.path.join(output_dir, cid))
#os.mkdir(os.path.join(output_dir, cid))
for seqid in raw[1:]:
otu[seqid] = cid
print >> sys.stderr, "finished reading", otu_filename
for cid in cids:
fa_d[cid] = open(os.path.join(output_dir, cid, cid + '.fasta'), 'w')
bw_d[cid] = BowTieWriter(os.path.join(output_dir, cid,
cid + '.bowtie'),
mode='w')
for r in SeqIO.parse(open(fasta_filename), 'fasta'):
if r.id not in otu:
continue
cid = otu[r.id]
fa_d[cid].write(">{0}\n{1}\n".format(r.id, r.seq))
for r in BowTieReader(bowtie_gz_filename, False):
try:
cid = otu[r['ID'].split()[0]]
except KeyError:
continue
bw_d[cid].write(r)
for handle in fa_d.itervalues():
handle.close()
for handle in bw_d.itervalues():
handle.close()
os.system("gzip " + handle.f.name)
def filter_low_qual_seqs(gz_filename1, gz_filename2, phred_offset,
phred_cutoff):
"""
Takes a BowTie-style gzipped file (ex: .aligned.composite.gz)
and retain only seqs that have every base phred >= <cutoff>
Outputs: .phred<cutoff>_passed for both files
"""
assert phred_offset >= 0
assert phred_cutoff >= 0
bad = 0
good = 0
start_t = time.time()
print gz_filename1, gz_filename2
f1 = BowTieWriter(gz_filename1 + ".phred{0}_passed".format(phred_cutoff),
'w')
f2 = BowTieWriter(gz_filename2 + ".phred{0}_passed".format(phred_cutoff),
'w')
for r1, r2 in itertools.izip(BowTieReader(gz_filename1, False),
BowTieReader(gz_filename2, False)):
if all(ord(x)-phred_offset >= phred_cutoff for x in r1['qual']) and \
all(ord(x)-phred_offset >= phred_cutoff for x in r2['qual']):
good += 1
f1.write(r1)
f2.write(r2)
else:
bad += 1
with open(gz_filename1 + ".phred{0}_passed.log".format(phred_cutoff),
'w') as f:
f.write(
"Running filter_low_qual_seq took {0} secs\n".format(time.time() -
start_t))
f.write("Input: " + gz_filename1 + ',' + gz_filename2 + '\n')
f.write("PhredCutoff: " + str(phred_cutoff) + '\n')
f.write("RemovedDueToLowQual: " + str(bad) + '\n')
f.write("RemainingTotal: " + str(good) + '\n')
def detect_primers_PE(input1, input2, output_prefix, f_primer, r_primer, min_match_len, max_mm, max_de, max_in):
"""
NOTE: this is for paired end reads that comes in two separate files
ex: DS19342_CTTGTA_L006_R1_001.fastq.gz and DS19342_CTTGTA_L006_R2_001.fastq.gz
Given a pair of reads from input1, input2:
1. Detect that F primer exists in one read and R primer in the other
2. If both reads pass primer detection, output
3. Otherwise, discard
Output: <output_prefix>.{F|R}primer_good
<output_prefix>.primer.bad
<output_prefix>.primer.log
"""
def process_primer(r, match_len, is_reverse):
# get record into miscBowTie.BowTieReader format
# strip away primers from seq & qual, properly rev comp!
r['offset'] = match_len
r['seq'] = r['seq'][match_len:]
r['qual'] = r['qual'][match_len:]
r['ref'] = 'NA'
if is_reverse:
r['seq'] = Seq(r['seq']).reverse_complement().tostring()
r['qual'] = r['qual'][::-1]
os.system("rm {0}.*primer_*".format(output_prefix))
Fgood = BowTieWriter(output_prefix + '.Fprimer_good')
Rgood = BowTieWriter(output_prefix + '.Rprimer_good')
hbad1 = FastqWriter(output_prefix + '.primer_bad.1')
hbad2 = FastqWriter(output_prefix + '.primer_bad.2')
hverbose = open(output_prefix + '.primer.verbose', 'w')
hlog = open(output_prefix + '.primer.log', 'w')
start_t = time.time()
good, bad = 0,0
pmF = PrimerMatch(f_primer)
pmR = PrimerMatch(r_primer)
for r1, r2 in itertools.izip(FastqReader(input1), FastqReader(input2)):
# NOTE: in the case of PE reads
# regardless of whether we're matching for F or R primer
# they would all appear at the 5' end of the read
# which is why we call match_primer_len with is_reverse = False
match_f_len1, mmf1 = match_primer_len(r1['seq'], f_primer, max_mm, min_match_len, False)
match_r_len1, mmr1 = match_primer_len(r1['seq'], r_primer, max_mm, min_match_len, False)
match_f_len2, mmf2 = match_primer_len(r2['seq'], f_primer, max_mm, min_match_len, False)
match_r_len2, mmr2 = match_primer_len(r2['seq'], r_primer, max_mm, min_match_len, False)
#match_f_len1 = match_f_len2 =match_r_len1=match_r_len2=0
if match_f_len1 > 0 and match_r_len2 > 0:
# case 1, read 1 is F, read 2 is R
good += 1
process_primer(r1, match_f_len1, False)
Fgood.write(r1)
process_primer(r2, match_r_len2, False)
Rgood.write(r2)
elif match_f_len2 > 0 and match_r_len1 > 0:
# case 2, read 1 is R, case 2 is F
good += 1
process_primer(r2, match_f_len2, False)
Fgood.write(r2)
process_primer(r1, match_r_len1, False)
Rgood.write(r1)
else:
pmF.make_suffix(r1['seq'])
pmF.match(min_match_len, max_mm, max_in, max_de)
if pmF.match_result is not None:
pmR.make_suffix(r2['seq'])
pmR.match(min_match_len, max_mm, max_in, max_de)
if pmR.match_result is not None: # case 1, read 1 is F, read 2 is R
good += 1
process_primer(r1, pmF.match_result.match_len, False)
Fgood.write(r1)
hverbose.write("{0}\t{1}\t{2}\n".format(r1['ID'], pmF.match_result.match_len, pmF.match_result.miss))
process_primer(r2, pmR.match_result.match_len, False)
Rgood.write(r2)
hverbose.write("{0}\t{1}\t{2}\n".format(r2['ID'], pmR.match_result.match_len, pmR.match_result.miss))
else:
hbad1.write(r1)
hbad2.write(r2)
bad += 1
else:
pmR.make_suffix(r1['seq'])
pmR.match(min_match_len, max_mm, max_in, max_de)
if pmR.match_result is not None:
pmF.make_suffix(r2['seq'])
pmF.match(min_match_len, max_mm, max_in, max_de)
if pmF.match_result is not None:
good += 1
# case 2, read 1 is R, read 2 is F
process_primer(r2, pmF.match_result.match_len, False)
hverbose.write("{0}\t{1}\t{2}\n".format(r2['ID'], pmF.match_result.match_len, pmF.match_result.miss))
Fgood.write(r2)
process_primer(r1, pmR.match_result.match_len, False)
Rgood.write(r1)
hverbose.write("{0}\t{1}\t{2}\n".format(r1['ID'], pmR.match_result.match_len, pmR.match_result.miss))
else:
# case 3: unresolved, bad read pair
hbad1.write(r1)
hbad2.write(r2)
bad += 1
hlog.write("Input 1: {0}\nInput 2: {1}\n".format(input1, input2))
hlog.write("F primer: {0}\nR primer: {1}\n".format(f_primer, r_primer))
hlog.write("Min match len: {0}\n".format(min_match_len))
hlog.write("Max mismatch: {0}\n".format(max_mm))
hlog.write("Max deletion: {0}\n".format(max_de))
hlog.write("Max insertion: {0}\n".format(max_in))
hlog.write("Primer detection and removal took {0} sec.\n".format(time.time()-start_t))
hlog.write("# of original reads: {0}\n".format(good+bad))
hlog.write("# of reads removed: {0} ({1:.2f})\n".format(bad,bad*1./(good+bad)))
hlog.write("# of reads remaining: {0} ({1:.2f})\n".format(good,good*1./(good+bad)))
Fgood.close()
Rgood.close()
hbad1.close()
hbad2.close()
hlog.close()
hverbose.close()
os.system("gzip " + Fgood.f.name)
os.system("gzip " + Rgood.f.name)
os.system("gzip " + hbad1.f.name)
os.system("gzip " + hbad2.f.name)
os.system("gzip " + hverbose.name)
qual += chr(int(-10*log10(e)+33))
seq += r2['seq'][N-delta:]
qual += r2['qual'][N-delta:]
return seq, qual, N-delta
if __name__ == "__main__":
from miscBowTie import BowTieReader, BowTieWriter
from cPickle import *
from optparse import OptionParser
parser = OptionParser()
parser.add_option("--input", dest="input", help="Input bowtie aligned file (gzipped)")
parser.add_option("--output", dest="output", help="Output composite read filename")
options, args = parser.parse_args()
reader = BowTieReader(options.input, is_paired=True)
print >> sys.stderr, "calculating base frequencies"
base_freq_pickle = options.input + ".base_freq.pickle"
if os.path.exists(base_freq_pickle):
base_freq = load(open(base_freq_pickle))
else:
base_freq = reader.get_base_frequency()
with open(options.input + ".base_freq.pickle", 'w') as f:
dump(base_freq, f)
print >> sys.stderr, "reading bowtie aligned file..."
writer = BowTieWriter(options.output)
for r1, r2 in reader:
seq, qual, overlap = compose2(r1, r2, base_freq)
writer.write_composite(r1, r2, seq, qual, overlap)
def remove_high_expected_error_PE(file1, file2, max_expected_error):
"""
Remove all reads where the expected error (sum of err probs from phred scores)
exceeds <max_expected_error>
"""
assert os.path.exists(file1) and os.path.exists(file2)
os.system("rm {0}.experror_*".format(file1))
os.system("rm {0}.experror_*".format(file2))
hgood1 = BowTieWriter(file1 + '.experror_good')
hgood2 = BowTieWriter(file2 + '.experror_good')
hbad1 = BowTieWriter(file1 + '.experror_bad')
hbad2 = BowTieWriter(file2 + '.experror_bad')
hlog = open(file1 + '.experror.log', 'w')
start_t = time.time()
good, bad = 0,0
for r1, r2 in itertools.izip(BowTieReader(file1, False), BowTieReader(file2, False)):
if sum(10**-((ord(x)-33)/10.) for x in r1['qual']) <= max_expected_error and \
sum(10**-((ord(x)-33)/10.) for x in r2['qual']) <= max_expected_error:
hgood1.write(r1)
hgood2.write(r2)
good += 1
else:
hbad1.write(r1)
hbad2.write(r2)
bad += 1
hlog.write("Expected error filtering took {0} sec.\n".format(time.time()-start_t))
hlog.write("Max allowed expected error: {0}\n".format(max_expected_error))
hlog.write("# of original reads: {0}\n".format(good+bad))
hlog.write("# of reads removed: {0} ({1:.2f})\n".format(bad,bad*1./(good+bad)))
hlog.write("# of reads remaining: {0} ({1:.2f})\n".format(good,good*1./(good+bad)))
hgood1.close()
hgood2.close()
hbad1.close()
hbad2.close()
hlog.close()
os.system("gzip " + hgood1.f.name)
os.system("gzip " + hgood2.f.name)
os.system("gzip " + hbad1.f.name)
os.system("gzip " + hbad2.f.name)
def remove_high_expected_error_PE(file1, file2, max_expected_error):
"""
Remove all reads where the expected error (sum of err probs from phred scores)
exceeds <max_expected_error>
"""
assert os.path.exists(file1) and os.path.exists(file2)
os.system("rm {0}.experror_*".format(file1))
os.system("rm {0}.experror_*".format(file2))
hgood1 = BowTieWriter(file1 + '.experror_good')
hgood2 = BowTieWriter(file2 + '.experror_good')
hbad1 = BowTieWriter(file1 + '.experror_bad')
hbad2 = BowTieWriter(file2 + '.experror_bad')
hlog = open(file1 + '.experror.log', 'w')
start_t = time.time()
good, bad = 0, 0
for r1, r2 in itertools.izip(BowTieReader(file1, False),
BowTieReader(file2, False)):
if sum(10**-((ord(x)-33)/10.) for x in r1['qual']) <= max_expected_error and \
sum(10**-((ord(x)-33)/10.) for x in r2['qual']) <= max_expected_error:
hgood1.write(r1)
hgood2.write(r2)
good += 1
else:
hbad1.write(r1)
hbad2.write(r2)
bad += 1
hlog.write("Expected error filtering took {0} sec.\n".format(time.time() -
start_t))
hlog.write("Max allowed expected error: {0}\n".format(max_expected_error))
hlog.write("# of original reads: {0}\n".format(good + bad))
hlog.write("# of reads removed: {0} ({1:.2f})\n".format(
bad, bad * 1. / (good + bad)))
hlog.write("# of reads remaining: {0} ({1:.2f})\n".format(
good, good * 1. / (good + bad)))
hgood1.close()
hgood2.close()
hbad1.close()
hbad2.close()
hlog.close()
os.system("gzip " + hgood1.f.name)
os.system("gzip " + hgood2.f.name)
os.system("gzip " + hbad1.f.name)
os.system("gzip " + hbad2.f.name)