def run(self):
assert self.reads or self.pairs or self.interleaved, 'No reads given'
io.check_name_uniqueness(self.reads, self.pairs, self.interleaved)
working = self.get_workspace()
working.setup_reference(self.references, bowtie=True)
working.update_param(snp_cost=2.0)
reference = working.get_reference()
log_file = open(self.log_filename(), 'wb')
with workspace.tempspace(dir=working.working_dir) as temp:
n = [0]
def tempname():
n[0] += 1
return temp / ('%d.fq' % n[0])
def convert(filename):
info = io.get_file_info(filename)
ok = selection.matches(
'type-fastq:[compression-none/compression-gzip/compression-bzip2]',
info)
if ok:
return filename
result_name = tempname()
with open(result_name, 'wb') as f:
for name, seq, qual in io.read_sequences(
filename, qualities='required'):
io.write_fastq(f, name, seq, qual)
return result_name
ones = []
twos = []
singles = []
for pair in self.pairs:
assert len(
pair) == 2, 'Need two files in each "pair:" section.'
ones.append(convert(pair[0]))
twos.append(convert(pair[1]))
for item in self.interleaved:
left_name = tempname()
right_name = tempname()
ones.append(left_name)
twos.append(right_name)
with open(left_name,'wb') as left, \
open(right_name,'wb') as right:
reader = io.read_sequences(item, qualities='required')
while True:
try:
name, seq, qual = reader.next()
except StopIteration:
break
io.write_fastq(left, name, seq, qual)
try:
name, seq, qual = reader.next()
except StopIteration:
raise grace.Error(
'Interleaved file contains odd number of sequences'
)
io.write_fastq(right, name, seq, qual)
for item in self.reads:
singles.append(convert(item))
cores = min(self.cores, legion.coordinator().get_cores())
command = ([
'bowtie2',
'--threads',
str(cores),
'--rg-id',
'1',
'--rg',
'SM:' + working.name,
] + self.bowtie_options +
['-x', reference.get_bowtie_index_prefix()])
commands = []
if ones:
commands.append(command +
['-1', ','.join(ones), '-2', ','.join(twos)])
if singles:
commands.append(command + ['-U', ','.join(singles)])
temp_bam_name = temp / 'temp.bam'
with io.pipe_to(['samtools', 'view', '-S', '-b', '-'],
stdout=open(temp_bam_name, 'wb'),
stderr=log_file) as f:
header_sent = False
for command in commands:
self.log.log('Running:\n' + ' '.join(command) + '\n')
with io.pipe_from(command, stderr=log_file,
cores=cores) as f_out:
for line in f_out:
if not header_sent or not line.startswith('@'):
f.write(line)
header_sent = True
#io.execute([
# 'samtools', 'sort', '-n', temp_bam_name, working/'alignments'
# ])
sam.sort_bam(temp_bam_name,
working / 'alignments',
by_name=True,
cores=self.cores)
log_file.close()
def run(self):
assert self.reads or self.pairs or self.interleaved, 'No reads given'
io.check_name_uniqueness(self.reads, self.pairs, self.interleaved)
working = self.get_workspace()
working.setup_reference(self.references, bowtie=True)
working.update_param(snp_cost=2.0)
reference = working.get_reference()
log_file = open(self.log_filename(),'wb')
with workspace.tempspace(dir=working.working_dir) as temp:
n = [ 0 ]
def tempname():
n[0] += 1
return temp/('%d.fq'%n[0])
def convert(filename):
info = io.get_file_info(filename)
ok = selection.matches('type-fastq:[compression-none/compression-gzip/compression-bzip2]', info)
if ok:
return filename
result_name = tempname()
with open(result_name,'wb') as f:
for name, seq, qual in io.read_sequences(filename, qualities='required'):
io.write_fastq(f, name, seq, qual)
return result_name
ones = [ ]
twos = [ ]
singles = [ ]
for pair in self.pairs:
assert len(pair) == 2, 'Need two files in each "pair:" section.'
ones.append(convert(pair[0]))
twos.append(convert(pair[1]))
for item in self.interleaved:
left_name = tempname()
right_name = tempname()
ones.append(left_name)
twos.append(right_name)
with open(left_name,'wb') as left, \
open(right_name,'wb') as right:
reader = io.read_sequences(item, qualities='required')
while True:
try:
name, seq, qual = reader.next()
except StopIteration:
break
io.write_fastq(left, name,seq,qual)
try:
name, seq, qual = reader.next()
except StopIteration:
raise grace.Error('Interleaved file contains odd number of sequences')
io.write_fastq(right, name,seq,qual)
for item in self.reads:
singles.append(convert(item))
cores = min(self.cores, legion.coordinator().get_cores())
command = (
[ 'bowtie2',
'--threads', str(cores),
'--rg-id', '1',
'--rg', 'SM:'+working.name,
] +
self.bowtie_options +
[ '-x', reference.get_bowtie_index_prefix() ]
)
commands = [ ]
if ones:
commands.append(command + [ '-1', ','.join(ones), '-2', ','.join(twos) ])
if singles:
commands.append(command + [ '-U', ','.join(singles) ])
temp_bam_name = temp/'temp.bam'
with io.pipe_to(
['samtools', 'view', '-S', '-b', '-'],
stdout=open(temp_bam_name,'wb'),
stderr=log_file
) as f:
header_sent = False
for command in commands:
self.log.log('Running:\n' + ' '.join(command) + '\n')
with io.pipe_from(
command,
stderr=log_file,
cores=cores
) as f_out:
for line in f_out:
if not header_sent or not line.startswith('@'):
f.write(line)
header_sent = True
#io.execute([
# 'samtools', 'sort', '-n', temp_bam_name, working/'alignments'
# ])
sam.sort_bam(temp_bam_name, working/'alignments', by_name=True, cores=self.cores)
log_file.close()
def run(self):
grace.require_shrimp_2()
grace.require_samtools()
assert self.references, 'No reference sequences given'
assert self.reads or self.pairs or self.interleaved, 'No reads given'
for pair in self.pairs:
assert len(pair) == 2, 'Two files required in each pair: section'
io.check_name_uniqueness(self.reads, self.pairs, self.interleaved)
read_sets = []
for item in self.reads:
read_sets.append(([item], False))
for item in self.pairs:
read_sets.append((item, True))
for item in self.interleaved:
read_sets.append(([item], True))
#Create working directory
workspace = self.get_workspace()
workspace.setup_reference(self.references)
workspace.update_param(snp_cost=25)
reference = workspace.get_reference()
reference_filename = reference.reference_fasta_filename()
cores = min(self.cores, legion.coordinator().get_cores())
default_options = {
'-E': None,
'-T': None,
'-N': str(cores),
'-n': '2',
'-w': '200%',
'-p': 'opp-in',
'-I': '0,500',
'-X': None,
}
if self.sam_unaligned:
default_options['--sam-unaligned'] = None
if self.half_paired:
default_options['--half-paired'] = None
else:
default_options['--no-half-paired'] = None
cutoff = '55%' #Default changed in SHRiMP 2.0.2
if '-h' in self.shrimp_options:
cutoff = self.shrimp_options[self.shrimp_options.index('-h') + 1]
#Run shrimp
bam_filename = io.abspath(self.output_dir, 'alignments.bam')
bam_prefix = io.abspath(self.output_dir, 'alignments')
bam_sorted_prefix = io.abspath(self.output_dir, 'alignments_sorted')
temp_filename = io.abspath(self.output_dir, 'temp.bam')
log_filename = io.abspath(self.output_dir, 'shrimp_log.txt')
log_file = open(log_filename, 'wb')
sam_eater = sam.Bam_writer(temp_filename)
sam_header_sent = [False]
n_seen = [0]
def eat(f):
for line in f:
if line.startswith('@'):
if sam_header_sent[0]: continue
else:
n_seen[0] += 1
if n_seen[0] % 100000 == 0:
grace.status('%s alignments produced' %
grace.pretty_number(n_seen[0]))
sam_eater.write_raw(line)
sam_header_sent[0] = True
def remove_pair_options(options):
for flag in ['-p', '-I']:
while flag in options:
pos = options.index(flag)
options = options[:pos] + options[pos + 2:]
for flag in ['--half-paired']:
while flag in options:
pos = options.index(flag)
options = options[:pos] + options[pos + 1:]
return options
for i, (filenames, is_paired) in enumerate(read_sets):
options = self.shrimp_options[:]
has_qualities = all(
len(io.read_sequences(filename, qualities=True).next()) ==
3 #A little ugly
for filename in filenames)
if has_qualities:
options.append('--fastq')
if len(filenames) == 1:
reads_parameters = [filenames[0]]
else:
reads_parameters = ['-1', filenames[0], '-2', filenames[1]]
if '--qv-offset' not in self.shrimp_options:
#guesses = [ ]
#for filename in filenames:
# guesses.append(io.guess_quality_offset(filename))
#assert len(set(guesses)) == 1, 'Conflicting quality offset guesses, please specify --qv-offset manually.'
#default_options['--qv-offset'] = str(guesses[0])
default_options['--qv-offset'] = str(
io.guess_quality_offset(*filenames))
default_options['--read-group'] = '%s,%s' % (
workspace.name.replace(',',
'_'), workspace.name.replace(',', '_'))
for flag in default_options:
if flag not in options:
options.append(flag)
if default_options[flag] is not None:
options.append(default_options[flag])
if not is_paired:
options = remove_pair_options(options)
grace.status('')
full_param = reference.shrimp_command(self.cs,
options + reads_parameters)
print >> sys.stderr, 'Running', ' '.join(full_param)
with io.pipe_from(full_param, stderr=log_file, cores=cores) as f:
eat(f)
log_file.close()
sam_eater.close()
grace.status('Sort')
#io.execute([
# 'samtools', 'sort', '-n', temp_filename, bam_prefix
#])
sam.sort_bam(temp_filename, bam_prefix, by_name=True, cores=self.cores)
os.unlink(temp_filename)
grace.status('')
def run(self):
grace.require_shrimp_2()
grace.require_samtools()
assert self.references, "No reference sequences given"
assert self.reads or self.pairs or self.interleaved, "No reads given"
for pair in self.pairs:
assert len(pair) == 2, "Two files required in each pair: section"
io.check_name_uniqueness(self.reads, self.pairs, self.interleaved)
read_sets = []
for item in self.reads:
read_sets.append(([item], False))
for item in self.pairs:
read_sets.append((item, True))
for item in self.interleaved:
read_sets.append(([item], True))
# Create working directory
workspace = self.get_workspace()
workspace.setup_reference(self.references)
workspace.update_param(snp_cost=25)
reference = workspace.get_reference()
reference_filename = reference.reference_fasta_filename()
cores = min(self.cores, legion.coordinator().get_cores())
default_options = {
"-E": None,
"-T": None,
"-N": str(cores),
"-n": "2",
"-w": "200%",
"-p": "opp-in",
"-I": "0,500",
"-X": None,
}
if self.sam_unaligned:
default_options["--sam-unaligned"] = None
if self.half_paired:
default_options["--half-paired"] = None
else:
default_options["--no-half-paired"] = None
cutoff = "55%" # Default changed in SHRiMP 2.0.2
if "-h" in self.shrimp_options:
cutoff = self.shrimp_options[self.shrimp_options.index("-h") + 1]
# Run shrimp
bam_filename = io.abspath(self.output_dir, "alignments.bam")
bam_prefix = io.abspath(self.output_dir, "alignments")
bam_sorted_prefix = io.abspath(self.output_dir, "alignments_sorted")
temp_filename = io.abspath(self.output_dir, "temp.bam")
log_filename = io.abspath(self.output_dir, "shrimp_log.txt")
log_file = open(log_filename, "wb")
sam_eater = sam.Bam_writer(temp_filename)
sam_header_sent = [False]
n_seen = [0]
def eat(f):
for line in f:
if line.startswith("@"):
if sam_header_sent[0]:
continue
else:
n_seen[0] += 1
if n_seen[0] % 100000 == 0:
grace.status("%s alignments produced" % grace.pretty_number(n_seen[0]))
sam_eater.write_raw(line)
sam_header_sent[0] = True
def remove_pair_options(options):
for flag in ["-p", "-I"]:
while flag in options:
pos = options.index(flag)
options = options[:pos] + options[pos + 2 :]
for flag in ["--half-paired"]:
while flag in options:
pos = options.index(flag)
options = options[:pos] + options[pos + 1 :]
return options
for i, (filenames, is_paired) in enumerate(read_sets):
options = self.shrimp_options[:]
has_qualities = all(
len(io.read_sequences(filename, qualities=True).next()) == 3 for filename in filenames # A little ugly
)
if has_qualities:
options.append("--fastq")
if len(filenames) == 1:
reads_parameters = [filenames[0]]
else:
reads_parameters = ["-1", filenames[0], "-2", filenames[1]]
if "--qv-offset" not in self.shrimp_options:
guesses = []
for filename in filenames:
guesses.append(io.guess_quality_offset(filename))
assert (
len(set(guesses)) == 1
), "Conflicting quality offset guesses, please specify --qv-offset manually."
default_options["--qv-offset"] = str(guesses[0])
default_options["--read-group"] = "%s,%s" % (
workspace.name.replace(",", "_"),
workspace.name.replace(",", "_"),
)
for flag in default_options:
if flag not in options:
options.append(flag)
if default_options[flag] is not None:
options.append(default_options[flag])
if not is_paired:
options = remove_pair_options(options)
grace.status("")
full_param = reference.shrimp_command(self.cs, options + reads_parameters)
print >>sys.stderr, "Running", " ".join(full_param)
with io.pipe_from(full_param, stderr=log_file, cores=cores) as f:
eat(f)
log_file.close()
sam_eater.close()
grace.status("Sort")
io.execute(["samtools", "sort", "-n", temp_filename, bam_prefix])
os.unlink(temp_filename)
grace.status("")
def run(self):
log = self.log
#quality_cutoff, args = grace.get_option_value(args, '--quality', int, 10)
#qoffset, args = grace.get_option_value(args, '--qoffset', int, None)
#clip_ambiguous, args = grace.get_option_value(args, '--clip-ambiguous', grace.as_bool, True)
#length_cutoff, args = grace.get_option_value(args, '--length', int, 24)
#adaptor_cutoff, args = grace.get_option_value(args, '--match', int, 10)
#max_error, args = grace.get_option_value(args, '--max-errors', int, 1)
#adaptor_set, args = grace.get_option_value(args, '--adaptors', str, 'truseq-adapter,truseq-srna,genomic,multiplexing,pe,srna')
#disallow_homopolymers, args = grace.get_option_value(args, '--homopolymers', grace.as_bool, False)
#reverse_complement, args = grace.get_option_value(args, '--revcom', grace.as_bool, False)
#trim_start, args = grace.get_option_value(args, '--trim-start', int, 0)
#trim_end, args = grace.get_option_value(args, '--trim-end', int, 0)
#output_fasta, args = grace.get_option_value(args, '--fasta', grace.as_bool, False)
#use_gzip, args = grace.get_option_value(args, '--gzip', grace.as_bool, True)
#output_rejects, args = grace.get_option_value(args, '--rejects', grace.as_bool, False)
#grace.expect_no_further_options(args)
prefix = self.prefix
log_name = os.path.split(prefix)[1]
quality_cutoff = self.quality
qoffset = self.qoffset
clip_ambiguous = self.clip_ambiguous
length_cutoff = self.length
adaptor_cutoff = self.match
max_error = self.max_errors
disallow_homopolymers = self.homopolymers
reverse_complement = self.revcom
trim_start = self.trim_start
trim_end = self.trim_end
output_fasta = self.fasta
use_gzip = self.gzip
output_rejects = self.rejects
iterators = []
filenames = []
any_paired = False
for filename in self.reads:
filenames.append(filename)
iterators.append(
itertools.izip(io.read_sequences(filename, qualities=True)))
for pair_filenames in self.pairs:
assert len(pair_filenames
) == 2, 'Expected a pair of files for "pairs" section.'
filenames.extend(pair_filenames)
any_paired = True
iterators.append(
itertools.izip(
io.read_sequences(pair_filenames[0], qualities=True),
io.read_sequences(pair_filenames[1], qualities=True)))
for filename in self.interleaved:
filenames.append(filename)
any_paired = True
iterators.append(
deinterleave(io.read_sequences(filename, qualities=True)))
fragment_reads = (2 if any_paired else 1)
read_in_fragment_names = ['read-1', 'read-2'
] if any_paired else ['read']
assert iterators, 'Nothing to clip'
io.check_name_uniqueness(self.reads, self.pairs, self.interleaved)
if qoffset is None:
#guesses = [ io.guess_quality_offset(filename) for filename in filenames ]
#assert len(set(guesses)) == 1, 'Conflicting quality offset guesses, please specify manually.'
#qoffset = guesses[0]
qoffset = io.guess_quality_offset(*filenames)
log.log('FASTQ offset seems to be %d\n' % qoffset)
quality_cutoff_char = chr(qoffset + quality_cutoff)
#log.log('Minimum quality: %d (%s)\n' % (quality_cutoff, quality_cutoff_char))
#log.log('Clip ambiguous bases: %s\n' % (grace.describe_bool(clip_ambiguous)))
#log.log('Minimum adaptor match: %d bases, %d errors\n' % (adaptor_cutoff, max_error))
#log.log('Minimum length: %d bases\n' % length_cutoff)
adaptor_seqs = []
adaptor_names = []
if self.adaptor_clip:
if self.adaptor_file:
adaptor_iter = io.read_sequences(self.adaptor_file)
else:
adaptor_iter = ADAPTORS
for name, seq in adaptor_iter:
seq = seq.upper().replace('U', 'T')
adaptor_seqs.append(seq)
adaptor_names.append(name)
adaptor_seqs.append(bio.reverse_complement(seq))
adaptor_names.append(name)
matcher = Matcher(adaptor_seqs, adaptor_names, max_error)
start_clips = [
collections.defaultdict(list) for i in xrange(fragment_reads)
]
end_clips = [
collections.defaultdict(list) for i in xrange(fragment_reads)
]
if output_fasta:
write_sequence = io.write_fasta_single_line
else:
write_sequence = io.write_fastq
f_single = io.open_possibly_compressed_writer(
self.reads_output_filenames()[0])
if fragment_reads == 2:
names = self.pairs_output_filenames(
)[0] if self.out_separate else self.interleaved_output_filenames()
f_paired = map(io.open_possibly_compressed_writer, names)
if output_rejects:
f_reject = io.open_possibly_compressed_writer(
self.rejects_output_filenames()[0])
n_single = 0
n_paired = 0
n_in_single = 0
n_in_paired = 0
total_in_length = [0] * fragment_reads
n_out = [0] * fragment_reads
n_q_clipped = [0] * fragment_reads
n_a_clipped = [0] * fragment_reads
n_homopolymers = [0] * fragment_reads
total_out_length = [0] * fragment_reads
#log.attach(open(prefix + '_log.txt', 'wb'))
for iterator in iterators:
for fragment in iterator:
if (n_in_single + n_in_paired) % 10000 == 0:
grace.status(
'Clipping fragment %s' %
grace.pretty_number(n_in_single + n_in_paired))
if len(fragment) == 1:
n_in_single += 1
else:
n_in_paired += 1
graduates = []
rejects = []
for i, (name, seq, qual) in enumerate(fragment):
seq = seq.upper()
total_in_length[i] += len(seq)
if self.trim_to:
seq = seq[:self.trim_to]
qual = qual[:self.trim_to]
start = trim_start
best_start = 0
best_len = 0
for j in xrange(len(seq) - trim_end):
if qual[j] < quality_cutoff_char or \
(clip_ambiguous and seq[j] not in 'ACGT'):
if best_len < j - start:
best_start = start
best_len = j - start
start = j + 1
j = len(seq) - trim_end
if best_len < j - start:
best_start = start
best_len = j - start
clipped_seq = seq[best_start:best_start + best_len]
clipped_qual = qual[best_start:best_start + best_len]
if len(clipped_seq) < length_cutoff:
n_q_clipped[i] += 1
rejects.append((name, seq, qual, 'quality'))
continue
match = matcher.match(clipped_seq)
if match and match[0] >= adaptor_cutoff:
clipped_seq = clipped_seq[match[0]:]
clipped_qual = clipped_qual[match[0]:]
start_clips[i][match[0]].append(match[1][0])
if len(clipped_seq) < length_cutoff:
n_a_clipped[i] += 1
rejects.append((name, seq, qual, 'adaptor'))
continue
match = matcher.match(bio.reverse_complement(clipped_seq))
if match and match[0] >= adaptor_cutoff:
clipped_seq = clipped_seq[:len(clipped_seq) - match[0]]
clipped_qual = clipped_qual[:len(clipped_qual) -
match[0]]
end_clips[i][match[0]].append(match[1][0])
if len(clipped_seq) < length_cutoff:
n_a_clipped[i] += 1
rejects.append((name, seq, qual, 'adaptor'))
continue
if disallow_homopolymers and len(set(clipped_seq)) <= 1:
n_homopolymers[i] += 1
rejects.append((name, seq, qual, 'homopolymer'))
continue
graduates.append((name, clipped_seq, clipped_qual))
n_out[i] += 1
total_out_length[i] += len(clipped_seq)
if output_rejects:
for name, seq, qual, reason in rejects:
write_sequence(f_reject, name + ' ' + reason, seq,
qual)
if graduates:
if reverse_complement:
graduates = [(name, bio.reverse_complement(seq),
qual[::-1])
for name, seq, qual in graduates]
if len(graduates) == 1:
n_single += 1
(name, seq, qual) = graduates[0]
write_sequence(f_single, name, seq, qual)
else:
assert len(graduates) == 2
n_paired += 1
# Write the pair to an interleaved file or separate l/r files
for (lr, (name, seq, qual)) in enumerate(graduates):
write_sequence(f_paired[lr % len(f_paired)], name,
seq, qual)
grace.status('')
if output_rejects:
f_reject.close()
if fragment_reads == 2:
map(lambda f: f.close(), f_paired)
f_single.close()
def summarize_clips(name, location, clips):
total = 0
for i in clips:
total += len(clips[i])
log.datum(log_name, name + ' adaptors clipped at ' + location,
total)
if not clips:
return
for i in xrange(min(clips), max(clips) + 1):
item = clips[i]
log.quietly_log('%3d bases: %10d ' % (i, len(item)))
if item:
avg_errors = float(sum(item2[0]
for item2 in item)) / len(item)
log.quietly_log(' avg errors: %5.2f ' % avg_errors)
counts = collections.defaultdict(int)
for item2 in item:
counts[item2[1]] += 1
#print counts
for no in sorted(counts,
key=lambda item2: counts[item2],
reverse=True)[:2]:
log.quietly_log('%dx%s ' %
(counts[no], matcher.names[no]))
if len(counts) > 2: log.quietly_log('...')
log.quietly_log('\n')
log.quietly_log('\n')
if n_in_paired:
log.datum(log_name, 'read-pairs', n_in_paired)
if n_in_single:
log.datum(log_name, 'single reads', n_in_single)
for i in xrange(fragment_reads):
if start_clips:
summarize_clips(read_in_fragment_names[i], 'start',
start_clips[i])
if end_clips:
summarize_clips(read_in_fragment_names[i], 'end', end_clips[i])
prefix = read_in_fragment_names[i]
log.datum(log_name, prefix + ' too short after quality clip',
n_q_clipped[i])
log.datum(log_name, prefix + ' too short after adaptor clip',
n_a_clipped[i])
if disallow_homopolymers:
log.datum(log_name, prefix + ' homopolymers',
n_homopolymers[i])
if fragment_reads > 1:
log.datum(log_name, prefix + ' kept', n_out[i])
log.datum(log_name, prefix + ' average input length',
float(total_in_length[i]) / (n_in_single + n_in_paired))
if n_out[i]:
log.datum(log_name, prefix + ' average output length',
float(total_out_length[i]) / n_out[i])
if fragment_reads == 2:
log.datum(log_name, 'pairs kept after clipping', n_paired)
log.datum(log_name, 'reads kept after clipping', n_single)
def run(self):
log = self.log
#quality_cutoff, args = grace.get_option_value(args, '--quality', int, 10)
#qoffset, args = grace.get_option_value(args, '--qoffset', int, None)
#clip_ambiguous, args = grace.get_option_value(args, '--clip-ambiguous', grace.as_bool, True)
#length_cutoff, args = grace.get_option_value(args, '--length', int, 24)
#adaptor_cutoff, args = grace.get_option_value(args, '--match', int, 10)
#max_error, args = grace.get_option_value(args, '--max-errors', int, 1)
#adaptor_set, args = grace.get_option_value(args, '--adaptors', str, 'truseq-adapter,truseq-srna,genomic,multiplexing,pe,srna')
#disallow_homopolymers, args = grace.get_option_value(args, '--homopolymers', grace.as_bool, False)
#reverse_complement, args = grace.get_option_value(args, '--revcom', grace.as_bool, False)
#trim_start, args = grace.get_option_value(args, '--trim-start', int, 0)
#trim_end, args = grace.get_option_value(args, '--trim-end', int, 0)
#output_fasta, args = grace.get_option_value(args, '--fasta', grace.as_bool, False)
#use_gzip, args = grace.get_option_value(args, '--gzip', grace.as_bool, True)
#output_rejects, args = grace.get_option_value(args, '--rejects', grace.as_bool, False)
#grace.expect_no_further_options(args)
prefix = self.prefix
log_name = os.path.split(prefix)[1]
quality_cutoff = self.quality
qoffset = self.qoffset
clip_ambiguous = self.clip_ambiguous
length_cutoff = self.length
adaptor_cutoff = self.match
max_error = self.max_errors
disallow_homopolymers = self.homopolymers
reverse_complement = self.revcom
trim_start = self.trim_start
trim_end = self.trim_end
output_fasta = self.fasta
use_gzip = self.gzip
output_rejects = self.rejects
iterators = [ ]
filenames = [ ]
any_paired = False
for filename in self.reads:
filenames.append(filename)
iterators.append(itertools.izip(
io.read_sequences(filename, qualities=True)
))
for pair_filenames in self.pairs:
assert len(pair_filenames) == 2, 'Expected a pair of files for "pairs" section.'
filenames.extend(pair_filenames)
any_paired = True
iterators.append(itertools.izip(
io.read_sequences(pair_filenames[0], qualities=True),
io.read_sequences(pair_filenames[1], qualities=True)
))
for filename in self.interleaved:
filenames.append(filename)
any_paired = True
iterators.append(deinterleave(
io.read_sequences(filename, qualities=True)
))
fragment_reads = (2 if any_paired else 1)
read_in_fragment_names = [ 'read-1', 'read-2' ] if any_paired else [ 'read' ]
assert iterators, 'Nothing to clip'
io.check_name_uniqueness(self.reads, self.pairs, self.interleaved)
if qoffset is None:
guesses = [ io.guess_quality_offset(filename) for filename in filenames ]
assert len(set(guesses)) == 1, 'Conflicting quality offset guesses, please specify manually.'
qoffset = guesses[0]
log.log('FASTQ offset seems to be %d\n' % qoffset)
quality_cutoff_char = chr(qoffset + quality_cutoff)
#log.log('Minimum quality: %d (%s)\n' % (quality_cutoff, quality_cutoff_char))
#log.log('Clip ambiguous bases: %s\n' % (grace.describe_bool(clip_ambiguous)))
#log.log('Minimum adaptor match: %d bases, %d errors\n' % (adaptor_cutoff, max_error))
#log.log('Minimum length: %d bases\n' % length_cutoff)
adaptor_seqs = [ ]
adaptor_names = [ ]
if self.adaptor_clip:
if self.adaptor_file:
adaptor_iter = io.read_sequences(self.adaptor_file)
else:
adaptor_iter = ADAPTORS
for name, seq in adaptor_iter:
seq = seq.upper().replace('U','T')
adaptor_seqs.append(seq)
adaptor_names.append(name)
adaptor_seqs.append(bio.reverse_complement(seq))
adaptor_names.append(name)
matcher = Matcher(adaptor_seqs, adaptor_names, max_error)
start_clips = [ collections.defaultdict(list) for i in xrange(fragment_reads) ]
end_clips = [ collections.defaultdict(list) for i in xrange(fragment_reads) ]
if output_fasta:
write_sequence = io.write_fasta_single_line
else:
write_sequence = io.write_fastq
f_single = io.open_possibly_compressed_writer(self.reads_output_filenames()[0])
if fragment_reads == 2:
names = self.pairs_output_filenames()[0] if self.out_separate else self.interleaved_output_filenames()
f_paired = map(io.open_possibly_compressed_writer, names)
if output_rejects:
f_reject = io.open_possibly_compressed_writer(self.rejects_output_filenames()[0])
n_single = 0
n_paired = 0
n_in_single = 0
n_in_paired = 0
total_in_length = [ 0 ] * fragment_reads
n_out = [ 0 ] * fragment_reads
n_q_clipped = [ 0 ] * fragment_reads
n_a_clipped = [ 0 ] * fragment_reads
n_homopolymers = [ 0 ] * fragment_reads
total_out_length = [ 0 ] * fragment_reads
#log.attach(open(prefix + '_log.txt', 'wb'))
for iterator in iterators:
for fragment in iterator:
if (n_in_single+n_in_paired) % 10000 == 0:
grace.status('Clipping fragment %s' % grace.pretty_number(n_in_single+n_in_paired))
if len(fragment) == 1:
n_in_single += 1
else:
n_in_paired += 1
graduates = [ ]
rejects = [ ]
for i, (name, seq, qual) in enumerate(fragment):
seq = seq.upper()
total_in_length[i] += len(seq)
if self.trim_to:
seq = seq[:self.trim_to]
qual = qual[:self.trim_to]
start = trim_start
best_start = 0
best_len = 0
for j in xrange(len(seq)-trim_end):
if qual[j] < quality_cutoff_char or \
(clip_ambiguous and seq[j] not in 'ACGT'):
if best_len < j-start:
best_start = start
best_len = j-start
start = j + 1
j = len(seq)-trim_end
if best_len < j-start:
best_start = start
best_len = j-start
clipped_seq = seq[best_start:best_start+best_len]
clipped_qual = qual[best_start:best_start+best_len]
if len(clipped_seq) < length_cutoff:
n_q_clipped[i] += 1
rejects.append( (name,seq,qual,'quality') )
continue
match = matcher.match(clipped_seq)
if match and match[0] >= adaptor_cutoff:
clipped_seq = clipped_seq[match[0]:]
clipped_qual = clipped_qual[match[0]:]
start_clips[i][match[0]].append( match[1][0] )
if len(clipped_seq) < length_cutoff:
n_a_clipped[i] += 1
rejects.append( (name,seq,qual,'adaptor') )
continue
match = matcher.match(bio.reverse_complement(clipped_seq))
if match and match[0] >= adaptor_cutoff:
clipped_seq = clipped_seq[: len(clipped_seq)-match[0] ]
clipped_qual = clipped_qual[: len(clipped_qual)-match[0] ]
end_clips[i][match[0]].append( match[1][0] )
if len(clipped_seq) < length_cutoff:
n_a_clipped[i] += 1
rejects.append( (name,seq,qual,'adaptor') )
continue
if disallow_homopolymers and len(set(clipped_seq)) <= 1:
n_homopolymers[i] += 1
rejects.append( (name,seq,qual,'homopolymer') )
continue
graduates.append( (name, clipped_seq, clipped_qual) )
n_out[i] += 1
total_out_length[i] += len(clipped_seq)
if output_rejects:
for name,seq,qual,reason in rejects:
write_sequence(f_reject, name + ' ' + reason, seq, qual)
if graduates:
if reverse_complement:
graduates = [
(name, bio.reverse_complement(seq), qual[::-1])
for name, seq, qual in graduates
]
if len(graduates) == 1:
n_single += 1
(name, seq, qual) = graduates[0]
write_sequence(f_single, name, seq, qual)
else:
assert len(graduates) == 2
n_paired += 1
# Write the pair to an interleaved file or separate l/r files
for (lr,(name, seq, qual)) in enumerate(graduates):
write_sequence(f_paired[lr%len(f_paired)], name, seq, qual)
grace.status('')
if output_rejects:
f_reject.close()
if fragment_reads == 2:
map(lambda f: f.close(), f_paired)
f_single.close()
def summarize_clips(name, location, clips):
total = 0
for i in clips:
total += len(clips[i])
log.datum(log_name, name + ' adaptors clipped at ' + location, total)
if not clips:
return
for i in xrange(min(clips), max(clips)+1):
item = clips[i]
log.quietly_log('%3d bases: %10d ' % (i, len(item)))
if item:
avg_errors = float(sum( item2[0] for item2 in item )) / len(item)
log.quietly_log(' avg errors: %5.2f ' % avg_errors)
counts = collections.defaultdict(int)
for item2 in item: counts[item2[1]] += 1
#print counts
for no in sorted(counts,key=lambda item2:counts[item2],reverse=True)[:2]:
log.quietly_log('%dx%s ' % (counts[no], matcher.names[no]))
if len(counts) > 2: log.quietly_log('...')
log.quietly_log('\n')
log.quietly_log('\n')
if n_in_paired:
log.datum(log_name,'read-pairs', n_in_paired)
if n_in_single:
log.datum(log_name,'single reads', n_in_single)
for i in xrange(fragment_reads):
if start_clips:
summarize_clips(read_in_fragment_names[i], 'start', start_clips[i])
if end_clips:
summarize_clips(read_in_fragment_names[i], 'end', end_clips[i])
prefix = read_in_fragment_names[i]
log.datum(log_name, prefix + ' too short after quality clip', n_q_clipped[i])
log.datum(log_name, prefix + ' too short after adaptor clip', n_a_clipped[i])
if disallow_homopolymers:
log.datum(log_name, prefix + ' homopolymers', n_homopolymers[i])
if fragment_reads > 1:
log.datum(log_name, prefix + ' kept', n_out[i])
log.datum(log_name, prefix + ' average input length', float(total_in_length[i]) / (n_in_single+n_in_paired))
if n_out[i]:
log.datum(log_name, prefix + ' average output length', float(total_out_length[i]) / n_out[i])
if fragment_reads == 2:
log.datum(log_name,'pairs kept after clipping', n_paired)
log.datum(log_name, 'reads kept after clipping', n_single)
def run(self):
grace.require_shrimp_2()
grace.require_samtools()
assert self.references, 'No reference sequences given'
assert self.reads or self.pairs or self.interleaved, 'No reads given'
for pair in self.pairs:
assert len(pair) == 2, 'Two files required in each pair: section'
io.check_name_uniqueness(self.reads, self.pairs, self.interleaved)
read_sets = [ ]
for item in self.reads:
read_sets.append( ([item], False) )
for item in self.pairs:
read_sets.append( (item, True) )
for item in self.interleaved:
read_sets.append( ([item], True) )
#Create working directory
workspace = self.get_workspace()
workspace.setup_reference(self.references)
workspace.update_param(snp_cost=25)
reference = workspace.get_reference()
reference_filename = reference.reference_fasta_filename()
cores = min(self.cores, legion.coordinator().get_cores())
default_options = {
'-E' : None,
'-T' : None,
'-N' : str(cores),
'-n':'2',
'-w':'200%',
'-p': 'opp-in',
'-I': '0,500',
'-X':None,
}
if self.sam_unaligned:
default_options['--sam-unaligned'] = None
if self.half_paired:
default_options['--half-paired'] = None
else:
default_options['--no-half-paired'] = None
cutoff = '55%' #Default changed in SHRiMP 2.0.2
if '-h' in self.shrimp_options:
cutoff = self.shrimp_options[ self.shrimp_options.index('-h')+1 ]
#Run shrimp
bam_filename = io.abspath(self.output_dir, 'alignments.bam')
bam_prefix = io.abspath(self.output_dir, 'alignments')
bam_sorted_prefix = io.abspath(self.output_dir, 'alignments_sorted')
temp_filename = io.abspath(self.output_dir, 'temp.bam')
log_filename = io.abspath(self.output_dir, 'shrimp_log.txt')
log_file = open(log_filename, 'wb')
sam_eater = sam.Bam_writer(temp_filename)
sam_header_sent = [False]
n_seen = [0]
def eat(f):
for line in f:
if line.startswith('@'):
if sam_header_sent[0]: continue
else:
n_seen[0] += 1
if n_seen[0] % 100000 == 0:
grace.status('%s alignments produced' % grace.pretty_number(n_seen[0]))
sam_eater.write_raw(line)
sam_header_sent[0] = True
def remove_pair_options(options):
for flag in ['-p','-I']:
while flag in options:
pos = options.index(flag)
options = options[:pos] + options[pos+2:]
for flag in ['--half-paired']:
while flag in options:
pos = options.index(flag)
options = options[:pos] + options[pos+1:]
return options
for i, (filenames, is_paired) in enumerate(read_sets):
options = self.shrimp_options[:]
has_qualities = all(
len( io.read_sequences(filename, qualities=True).next() ) == 3 #A little ugly
for filename in filenames
)
if has_qualities:
options.append( '--fastq' )
if len(filenames) == 1:
reads_parameters = [ filenames[0] ]
else:
reads_parameters = [ '-1', filenames[0], '-2', filenames[1] ]
if '--qv-offset' not in self.shrimp_options:
#guesses = [ ]
#for filename in filenames:
# guesses.append(io.guess_quality_offset(filename))
#assert len(set(guesses)) == 1, 'Conflicting quality offset guesses, please specify --qv-offset manually.'
#default_options['--qv-offset'] = str(guesses[0])
default_options['--qv-offset'] = str( io.guess_quality_offset(*filenames) )
default_options['--read-group'] = '%s,%s' % (
workspace.name.replace(',','_'),
workspace.name.replace(',','_')
)
for flag in default_options:
if flag not in options:
options.append(flag)
if default_options[flag] is not None:
options.append(default_options[flag])
if not is_paired:
options = remove_pair_options(options)
grace.status('')
full_param = reference.shrimp_command(self.cs, options + reads_parameters)
print >> sys.stderr, 'Running', ' '.join(full_param)
with io.pipe_from(full_param,
stderr=log_file,
cores=cores) as f:
eat(f)
log_file.close()
sam_eater.close()
grace.status('Sort')
#io.execute([
# 'samtools', 'sort', '-n', temp_filename, bam_prefix
#])
sam.sort_bam(temp_filename, bam_prefix, by_name=True, cores=self.cores)
os.unlink(temp_filename)
grace.status('')
