def main(args):
size, args = grace.get_option_value(args,'--size',int,200)
stride, args = grace.get_option_value(args,'--stride',int,50)
grace.expect_no_further_options(args)
if not args:
print USAGE
return 1
for filename in args:
for name, seq in io.read_sequences(filename):
name_parts = name.split(None, 1)
name = name_parts[0]
if len(name_parts) > 1:
desc = ' ' + name_parts[1]
else:
desc = ''
for i in xrange(-size+stride,len(seq),stride):
start = max(0,min(len(seq),i))
end = max(0,min(len(seq), i+size))
io.write_fasta(
sys.stdout,
'%s:%d..%d' % (name,start+1,end) + desc,
seq[start:end]
)
return 0
def normalize(args):
min_depth, args = grace.get_option_value(args, '--min-depth', int, 5)
grace.expect_no_further_options(args)
if len(args) < 2:
print NORMALIZE_HELP
raise grace.Help_shown()
dirnames = args
filenames = []
for dirname in dirnames:
assert os.path.isdir(dirname), dirname + ' is not a directory'
filenames.append(
sorted(
item for item in os.listdir(dirname)
#if item.endswith('.userplot') and not item.endswith('-norm.userplot')
if item.endswith('-depth.userplot')
and not item.endswith('-ambiguous-depth.userplot')
and not item.endswith('-pairspan-depth.userplot')))
for i in xrange(1, len(dirnames)):
if filenames[i] != filenames[0]:
raise grace.Error('Userplots in %s differ from those in %s' %
(dirnames[i], dirnames[0]))
filenames = filenames[0]
for filename in filenames:
normalize_files(dirnames, filename[:-15], min_depth)
def main(args):
size, args = grace.get_option_value(args, '--size', int, 200)
stride, args = grace.get_option_value(args, '--stride', int, 50)
grace.expect_no_further_options(args)
if not args:
print USAGE
return 1
for filename in args:
for name, seq in io.read_sequences(filename):
name_parts = name.split(None, 1)
name = name_parts[0]
if len(name_parts) > 1:
desc = ' ' + name_parts[1]
else:
desc = ''
for i in xrange(-size + stride, len(seq), stride):
start = max(0, min(len(seq), i))
end = max(0, min(len(seq), i + size))
io.write_fasta(sys.stdout,
'%s:%d..%d' % (name, start + 1, end) + desc,
seq[start:end])
return 0
def normalize(args):
min_depth, args = grace.get_option_value(args, '--min-depth', int, 5)
grace.expect_no_further_options(args)
if len(args) < 2:
print NORMALIZE_HELP
raise grace.Help_shown()
dirnames = args
filenames = [ ]
for dirname in dirnames:
assert os.path.isdir(dirname), dirname + ' is not a directory'
filenames.append(sorted(
item for item in os.listdir(dirname)
#if item.endswith('.userplot') and not item.endswith('-norm.userplot')
if item.endswith('-depth.userplot')
and not item.endswith('-ambiguous-depth.userplot')
and not item.endswith('-pairspan-depth.userplot')
))
for i in xrange(1,len(dirnames)):
if filenames[i] != filenames[0]:
raise grace.Error('Userplots in %s differ from those in %s' % (dirnames[i], dirnames[0]))
filenames = filenames[0]
for filename in filenames:
normalize_files(dirnames, filename[:-15], min_depth)
def import_(args):
grace.expect_no_further_options(args)
for item in args:
sample = Options()
options.samples.append(sample)
sample.imported = True
sample.clip_dest = None
sample.align_dest = absolutize(item)
def front_command(args):
grace.expect_no_further_options(args)
if len(args) < 1:
return
output_dir.append(args[0])
input_reference_filenames.extend(
[os.path.abspath(filename) for filename in args[1:]])
def front_command(args):
grace.expect_no_further_options(args)
if len(args) < 1:
return
output_dir.append(args[0])
input_reference_filenames.extend(
[ os.path.abspath(filename) for filename in args[1:] ])
def plot(args):
log_it, args = grace.get_option_value(args, '--log', grace.as_bool, False)
grace.expect_no_further_options(args)
import numpy, pylab
pylab.rcParams['axes.formatter.limits'] = [-20, 20]
pylab.figure(figsize=(10, 4))
maximum = 0
for filename in args:
parts = filename.split('~~', 1)
data = []
f = open(parts[0], 'rb')
for line in f:
data.append(float(line.strip()))
f.close()
data = numpy.array(data)
maximum = max(maximum, numpy.maximum.reduce(data))
#if log_it:
# data = numpy.log(data + 1.0) / numpy.log(2.0)
if log_it:
pylab.semilogy(numpy.arange(1,
len(data) + 1),
data,
label=parts[-1])
else:
pylab.plot(numpy.arange(1, len(data) + 1), data, label=parts[-1])
if len(args) > 1:
pylab.legend()
if log_it:
pylab.ylim((1, maximum**1.2))
else:
pylab.ylim((0, maximum * 1.2))
pylab.show()
def plot(args):
log_it, args = grace.get_option_value(args, '--log', grace.as_bool, False)
grace.expect_no_further_options(args)
import numpy, pylab
pylab.rcParams['axes.formatter.limits'] = [ -20, 20 ]
pylab.figure(figsize=(10,4))
maximum = 0
for filename in args:
parts = filename.split('~~', 1)
data = [ ]
f = open(parts[0],'rb')
for line in f:
data.append(float(line.strip()))
f.close()
data = numpy.array(data)
maximum = max(maximum,numpy.maximum.reduce(data))
#if log_it:
# data = numpy.log(data + 1.0) / numpy.log(2.0)
if log_it:
pylab.semilogy( numpy.arange(1,len(data)+1), data, label=parts[-1] )
else:
pylab.plot( numpy.arange(1,len(data)+1), data, label=parts[-1] )
if len(args) > 1:
pylab.legend()
if log_it:
pylab.ylim( (1,maximum**1.2) )
else:
pylab.ylim( (0,maximum*1.2) )
pylab.show()
def pairs_command(args):
grace.expect_no_further_options(args)
assert len(args) == 2, 'Expected exactly two files in "pairs"'
reads_filenames.append([ os.path.abspath(filename) for filename in args ])
def interleaved(args):
grace.expect_no_further_options(args)
sample.interleaved.extend(args)
def pairs(args):
grace.expect_no_further_options(args)
assert len(args) == 2, 'Expected exactly two files in "pairs:"'
sample.pairs.append(args)
def main(args):
default_transl_table, args = grace.get_option_value(
args, '--transl_table', int, 11)
use_coverage, args = grace.get_flag(args, '--use-coverage')
coverage_cutoff, args = grace.get_option_value(args, '--coverage-cutoff',
float, 0.1)
tabular, args = grace.get_flag(args, '--tabular')
noheader, args = grace.get_flag(args, '--noheader')
verbose, args = grace.get_flag(args, '--verbose')
bandwidth, args = grace.get_option_value(args, '--band', int, 20)
grace.expect_no_further_options(args)
if len(args) != 2:
print USAGE
return 1
genbank_filename = args[0]
alignment_filename = args[1]
if os.path.isdir(alignment_filename):
alignment_filename = os.path.join(alignment_filename, 'alignment.maf')
working_dir = os.path.split(alignment_filename)[0]
alignments = load_alignments(alignment_filename)
summaries = []
details = []
if not noheader:
fields = 'Sequence\tLocus tag\tOld length (aa)\tNew length (aa)\tAmino acid changes\t'
if use_coverage:
fields += 'Unambiguous coverage vs expected\t\tAmbiguous coverage vs expected\t\tAmbiguous percent with any hits\t'
fields += 'Gene\tProduct'
if tabular: fields += '\tChanges of note'
print fields
for record in SeqIO.parse(
io.open_possibly_compressed_file(genbank_filename), 'genbank'):
sequence = record.seq.tostring()
for name, seq1, seq2, alignment in alignments:
if seq1 == sequence: break
else:
raise grace.Error(
'Genbank record %s sequence not identical to any reference sequence'
% record.id)
if use_coverage:
depth = get_graph(working_dir, name, 'depth')
ambiguous_depth = get_graph(working_dir, name, 'ambiguous-depth')
median_depth = numpy.median(depth)
median_ambiguous_depth = numpy.median(ambiguous_depth)
ambiguous_factor = float(median_ambiguous_depth) / median_depth
depth_expect = expected_depth(name, sequence, depth,
ambiguous_depth)
for feature in record.features:
if feature.type != 'CDS': continue
if 'locus_tag' not in feature.qualifiers:
locus_tag = '%d..%d' % (feature.location.nofuzzy_start + 1,
feature.location.nofuzzy_end)
else:
locus_tag = feature.qualifiers['locus_tag'][0]
if 'transl_table' in feature.qualifiers:
transl_table_no = int(feature.qualifiers['transl_table'][0])
else:
assert default_transl_table is not None, 'No /transl_table for CDS, and default transl_table not given'
transl_table_no = default_transl_table
transl_table = CodonTable.ambiguous_dna_by_id[transl_table_no]
start_codons = transl_table.start_codons
try:
feature_alignment = alignment_from_feature(sequence, feature)
except Weird_alignment:
warn('%s has a location I could not handle, skipping, sorry' %
locus_tag)
continue
dna = []
new_dna = []
shifts = []
for i in xrange(feature_alignment.end2):
p1 = feature_alignment.back_project(i, left=False)
p2 = feature_alignment.back_project(i + 1, left=True)
assert abs(p2 - p1) < 2
dna.append(sequence_slice(sequence, p1, p2))
p1a = alignment.project(p1, left=False)
p2a = alignment.project(p2, left=False) #Hmm
diff = (p2 - p1) - (p2a - p1a)
#if diff:
# if diff%3:
# frame_shift = True
# else:
# frame_preserving_shift = True
new_dna.append(sequence_slice(seq2, p1a, p2a))
if diff:
shifts.append((i, dna[-1], new_dna[-1]))
dna = ''.join(dna)
new_dna = ''.join(new_dna)
# This usually indicated a CDS truncated at the start?
# in which case, will probably fail some way or other down the line.
if 'codon_start' in feature.qualifiers:
codon_start = int(feature.qualifiers['codon_start'][0]) - 1
else:
codon_start = 0
dna = dna[codon_start:]
new_dna = new_dna[codon_start:]
if len(dna) % 3 != 0:
warn(locus_tag + ' length not a multiple of 3')
#assert len(new_dna) % 3 == 0
protein = Seq.Seq(dna).translate(table=transl_table_no).tostring()
# http://en.wikipedia.org/wiki/Start_codon is always translated to M
protein = 'M' + protein[1:]
if dna[:3] not in start_codons:
warn(locus_tag + ' has unknown start codon: ' + dna[:3])
original_lacks_stop_codon = not protein.endswith('*')
if original_lacks_stop_codon:
warn(locus_tag + ' lacks end codon')
original_stops_before_end = '*' in protein[:-1]
if original_stops_before_end:
warn(locus_tag + ' contains stop codon before end')
if 'translation' in feature.qualifiers:
expect = feature.qualifiers['translation'][0]
if protein[:-1] != expect:
warn(
locus_tag +
' translation given in feature does not match translation from DNA'
)
new_protein = Seq.Seq(new_dna).translate(
table=transl_table_no).tostring()
new_protein = 'M' + new_protein[1:]
# If end codon changed, find new end
# Don't bother if there are unknown amino acids or
# the original protein lacks a stop codon
if 'X' not in new_protein and '*' not in new_protein and not original_lacks_stop_codon:
#This is very inefficient
i = feature_alignment.end2
while True:
p1 = feature_alignment.back_project(i, left=False)
p2 = feature_alignment.back_project(i + 1, left=True)
p1a = alignment.project(p1, left=False)
p2a = alignment.project(p2, left=False) #Hmm
if p1a < 0 or p2a < 0 or p1a > len(seq2) or p2a > len(
seq2):
break
new_dna += sequence_slice(seq2, p1a, p2a)
new_protein = Seq.Seq(new_dna).translate(
table=transl_table_no).tostring()
new_protein = 'M' + new_protein[1:]
if 'X' in new_protein or '*' in new_protein: break
i += 1
# Is the protein shorter?
# Don't bother checking if the original protein has extra stop codons
if '*' in new_protein and not original_stops_before_end:
new_protein = new_protein[:new_protein.index('*') + 1]
# If indels occurred, do an alignment
# Don't bother otherwise
if shifts:
# Penalize gaps with cost 2 (vs 1 for mismatch)
# If lengths don't match, pad with spaces (won't match longer seq),
# aligner prefers mismatch to gaps
#result = pairwise2.align.globalxs(protein + ' '*max(0,len(new_protein)-len(protein)),
# new_protein + ' '*max(0,len(protein)-len(new_protein)),
# -2.001,-2.000)[0]
# 2.001 : very slightly prefer contiguous gaps. Also much faster!
result = band_limited_align(
protein + ' ' * max(0,
len(new_protein) - len(protein)),
new_protein + ' ' * max(0,
len(protein) - len(new_protein)),
bandwidth)
protein_ali = result[0]
new_protein_ali = result[1]
else:
protein_ali = protein
new_protein_ali = new_protein
diffs = []
j = 0
k = 0
for i in xrange(min(len(new_protein_ali), len(protein_ali))):
if protein_ali[i] != ' ' and new_protein_ali[i] != ' ' and (
protein_ali[i] == '-' or new_protein_ali[i] == '-'
or not bio.might_be_same_amino(protein_ali[i],
new_protein_ali[i])):
diffs.append((i, j, k))
if protein_ali[i] != '-':
j += 1
if new_protein_ali[i] != '-':
k += 1
diff_start = not bio.might_be_same_base(new_dna[0],dna[0]) or \
not bio.might_be_same_base(new_dna[1],dna[1]) or \
not bio.might_be_same_base(new_dna[2],dna[2])
interesting_coverage = False
if use_coverage:
cds_depth = depth[feature_alignment.start1:
feature_alignment.end1] #/ median_depth
if not feature_alignment.forward1: cds_depth = cds_depth[::-1]
cds_ambiguous_depth = ambiguous_depth[
feature_alignment.start1:
feature_alignment.end1] #/ median_ambiguous_depth
if not feature_alignment.forward1:
cds_ambiguous_depth = cds_ambiguous_depth[::-1]
cds_depth_expect = depth_expect[feature_alignment.
start1:feature_alignment.end1]
if not feature_alignment.forward1:
cds_depth_expect = cds_depth_expect[::-1]
#cds_average_depth_ratio = numpy.average(depth[feature_alignment.start1:feature_alignment.end1]) / median_depth
#cds_average_ambiguous_depth_ratio = numpy.average(ambiguous_depth[feature_alignment.start1:feature_alignment.end1]) / median_ambiguous_depth
#line += '%.1f\t' % cds_average_depth_ratio
#line += '%.1f\t' % cds_average_ambiguous_depth_ratio
#line += '%.1f..%.1f\t' % (numpy.minimum.reduce(cds_depth)/median_depth, numpy.maximum.reduce(cds_depth)/median_depth)
#line += '%.1f+/-%.1f\t' % (numpy.average(cds_depth)/median_depth, numpy.var(cds_depth)**0.5/median_depth)
#line += '%.1f..%.1f\t' % (numpy.minimum.reduce(cds_ambiguous_depth)/median_ambiguous_depth, numpy.maximum.reduce(cds_ambiguous_depth)/median_ambiguous_depth)
avg_expect = numpy.average(cds_depth_expect)
if avg_expect > 0.0:
cds_avg_depth = numpy.average(cds_depth) / avg_expect
cds_avg_ambiguous_depth = numpy.average(
cds_ambiguous_depth) / avg_expect / ambiguous_factor
strange = ((cds_depth >= cds_depth_expect * 1.5) |
(cds_ambiguous_depth <= cds_depth_expect *
(0.5 * ambiguous_factor)))
interesting_coverage = numpy.average(
strange) >= coverage_cutoff
if interesting_coverage or diffs or diff_start or shifts or len(
new_protein) != len(protein):
line = name + '\t' + locus_tag + '\t' + \
'%d\t' % (len(protein)-1) + \
'%d\t' % (len(new_protein)-1) + \
'%d\t' % len(diffs)
if use_coverage:
if avg_expect <= 0.0:
line += '\t\t\t'
else:
line += '%.1f\t' % (cds_avg_depth) + graphlet(
cds_depth, cds_depth_expect) + '\t'
line += '%.1f\t' % (
cds_avg_ambiguous_depth) + graphlet(
cds_ambiguous_depth,
cds_depth_expect * ambiguous_factor) + '\t'
line += '%.1f%%\t' % (
numpy.average(cds_ambiguous_depth > 0.0) * 100.0)
line += '%s\t' % feature.qualifiers.get('gene',[''])[0] + \
'%s' % feature.qualifiers.get('product',[''])[0]
notes = []
if use_coverage and 'X' in new_protein:
xs = new_protein.count('X')
if xs == len(new_protein) - 1: #First is M, so len-1
notes.append('\ No consensus')
else:
notes.append('\ No consensus for %d aa' %
(new_protein.count('X')))
if len(new_protein) < len(protein):
notes.append('\ Shorter by %d aa' %
(len(protein) - len(new_protein)))
if len(new_protein) > len(protein):
notes.append('\ Longer by %d aa' %
(len(new_protein) - len(protein)))
if diff_start:
notes.append('\ Start changed: %s -> %s' %
(dna[:3], new_dna[:3]))
if new_dna[:3] not in start_codons:
notes.append(' No longer a start codon!')
if shifts:
notes.append('\ Indels:')
for pos, old, new in shifts:
notes.append(' base %5d / codon %5d %s -> %s' %
(pos + 1,
(pos // 3) + 1, old, new or '-'))
if diffs:
if verbose:
notes.append('\ Amino acid changes:')
for i, j, k in diffs:
notes.append(
' codon %5d %s->%s (%s->%s)' %
(j + 1, protein_ali[i], new_protein_ali[i],
dna[j * 3:j * 3 + 3] if protein_ali[i] != '-'
else '-', new_dna[k * 3:k * 3 + 3]
if new_protein_ali[i] != '-' else '-'))
#if len(new_protein) > len(protein):
# print 'New protein is longer:', new_protein[len(protein):]
#if len(new_protein) < len(protein):
# print 'New protein is shorter:', protein[len(new_protein):]
#print protein
#print new_protein
if tabular:
print line + '\t' + ' '.join(
[' '.join(note.strip().split()) for note in notes])
else:
print line
for note in notes:
print '\t' + note
return 0
def main(args):
default_transl_table, args = grace.get_option_value(args, '--transl_table', int, 11)
use_coverage, args = grace.get_flag(args, '--use-coverage')
coverage_cutoff, args = grace.get_option_value(args, '--coverage-cutoff', float, 0.1)
tabular, args = grace.get_flag(args, '--tabular')
noheader, args = grace.get_flag(args, '--noheader')
verbose, args = grace.get_flag(args, '--verbose')
bandwidth, args = grace.get_option_value(args, '--band', int, 20)
grace.expect_no_further_options(args)
if len(args) != 2:
print USAGE
return 1
genbank_filename = args[0]
alignment_filename = args[1]
if os.path.isdir(alignment_filename):
alignment_filename = os.path.join(alignment_filename, 'alignment.maf')
working_dir = os.path.split(alignment_filename)[0]
alignments = load_alignments(alignment_filename)
summaries = [ ]
details = [ ]
if not noheader:
fields = 'Sequence\tLocus tag\tOld length (aa)\tNew length (aa)\tAmino acid changes\t'
if use_coverage: fields += 'Unambiguous coverage vs expected\t\tAmbiguous coverage vs expected\t\tAmbiguous percent with any hits\t'
fields += 'Gene\tProduct'
if tabular: fields += '\tChanges of note'
print fields
for record in SeqIO.parse(io.open_possibly_compressed_file(genbank_filename),'genbank'):
sequence = record.seq.tostring()
for name, seq1, seq2, alignment in alignments:
if seq1 == sequence: break
else:
raise grace.Error('Genbank record %s sequence not identical to any reference sequence' % record.id)
if use_coverage:
depth = get_graph(working_dir, name, 'depth')
ambiguous_depth = get_graph(working_dir, name, 'ambiguous-depth')
median_depth = numpy.median(depth)
median_ambiguous_depth = numpy.median(ambiguous_depth)
ambiguous_factor = float(median_ambiguous_depth) / median_depth
depth_expect = expected_depth(name, sequence, depth, ambiguous_depth)
for feature in record.features:
if feature.type != 'CDS': continue
if 'locus_tag' not in feature.qualifiers:
locus_tag = '%d..%d' % (feature.location.nofuzzy_start+1,feature.location.nofuzzy_end)
else:
locus_tag = feature.qualifiers['locus_tag'][0]
if 'transl_table' in feature.qualifiers:
transl_table_no = int(feature.qualifiers['transl_table'][0])
else:
assert default_transl_table is not None, 'No /transl_table for CDS, and default transl_table not given'
transl_table_no = default_transl_table
transl_table = CodonTable.ambiguous_dna_by_id[transl_table_no]
start_codons = transl_table.start_codons
try:
feature_alignment = alignment_from_feature(sequence, feature)
except Weird_alignment:
warn('%s has a location I could not handle, skipping, sorry' % locus_tag)
continue
dna = [ ]
new_dna = [ ]
shifts = [ ]
for i in xrange(feature_alignment.end2):
p1 = feature_alignment.back_project(i, left=False)
p2 = feature_alignment.back_project(i+1, left=True)
assert abs(p2-p1) < 2
dna.append( sequence_slice(sequence,p1,p2) )
p1a = alignment.project(p1, left=False)
p2a = alignment.project(p2, left=False) #Hmm
diff = (p2-p1)-(p2a-p1a)
#if diff:
# if diff%3:
# frame_shift = True
# else:
# frame_preserving_shift = True
new_dna.append( sequence_slice(seq2,p1a,p2a) )
if diff:
shifts.append((i,dna[-1],new_dna[-1]))
dna = ''.join(dna)
new_dna = ''.join(new_dna)
# This usually indicated a CDS truncated at the start?
# in which case, will probably fail some way or other down the line.
if 'codon_start' in feature.qualifiers:
codon_start = int(feature.qualifiers['codon_start'][0]) - 1
else:
codon_start = 0
dna = dna[codon_start:]
new_dna = new_dna[codon_start:]
if len(dna) % 3 != 0:
warn(locus_tag + ' length not a multiple of 3')
#assert len(new_dna) % 3 == 0
protein = Seq.Seq(dna).translate(table=transl_table_no).tostring()
# http://en.wikipedia.org/wiki/Start_codon is always translated to M
protein = 'M' + protein[1:]
if dna[:3] not in start_codons:
warn(locus_tag + ' has unknown start codon: ' + dna[:3])
original_lacks_stop_codon = not protein.endswith('*')
if original_lacks_stop_codon:
warn(locus_tag + ' lacks end codon')
original_stops_before_end = '*' in protein[:-1]
if original_stops_before_end:
warn(locus_tag + ' contains stop codon before end')
if 'translation' in feature.qualifiers:
expect = feature.qualifiers['translation'][0]
if protein[:-1] != expect:
warn(locus_tag + ' translation given in feature does not match translation from DNA')
new_protein = Seq.Seq(new_dna).translate(table=transl_table_no).tostring()
new_protein = 'M' + new_protein[1:]
# If end codon changed, find new end
# Don't bother if there are unknown amino acids or
# the original protein lacks a stop codon
if 'X' not in new_protein and '*' not in new_protein and not original_lacks_stop_codon:
#This is very inefficient
i = feature_alignment.end2
while True:
p1 = feature_alignment.back_project(i, left=False)
p2 = feature_alignment.back_project(i+1, left=True)
p1a = alignment.project(p1, left=False)
p2a = alignment.project(p2, left=False) #Hmm
if p1a < 0 or p2a < 0 or p1a > len(seq2) or p2a > len(seq2):
break
new_dna += sequence_slice(seq2,p1a,p2a)
new_protein = Seq.Seq(new_dna).translate(table=transl_table_no).tostring()
new_protein = 'M' + new_protein[1:]
if 'X' in new_protein or '*' in new_protein: break
i += 1
# Is the protein shorter?
# Don't bother checking if the original protein has extra stop codons
if '*' in new_protein and not original_stops_before_end:
new_protein = new_protein[:new_protein.index('*')+1]
# If indels occurred, do an alignment
# Don't bother otherwise
if shifts:
# Penalize gaps with cost 2 (vs 1 for mismatch)
# If lengths don't match, pad with spaces (won't match longer seq),
# aligner prefers mismatch to gaps
#result = pairwise2.align.globalxs(protein + ' '*max(0,len(new_protein)-len(protein)),
# new_protein + ' '*max(0,len(protein)-len(new_protein)),
# -2.001,-2.000)[0]
# 2.001 : very slightly prefer contiguous gaps. Also much faster!
result = band_limited_align(protein + ' '*max(0,len(new_protein)-len(protein)),
new_protein + ' '*max(0,len(protein)-len(new_protein)),
bandwidth)
protein_ali = result[0]
new_protein_ali = result[1]
else:
protein_ali = protein
new_protein_ali = new_protein
diffs = [ ]
j = 0
k = 0
for i in xrange(min(len(new_protein_ali),len(protein_ali))):
if protein_ali[i] != ' ' and new_protein_ali[i] != ' ' and (
protein_ali[i] == '-' or
new_protein_ali[i] == '-' or
not bio.might_be_same_amino(protein_ali[i], new_protein_ali[i]) ):
diffs.append((i,j,k))
if protein_ali[i] != '-':
j += 1
if new_protein_ali[i] != '-':
k += 1
diff_start = not bio.might_be_same_base(new_dna[0],dna[0]) or \
not bio.might_be_same_base(new_dna[1],dna[1]) or \
not bio.might_be_same_base(new_dna[2],dna[2])
interesting_coverage = False
if use_coverage:
cds_depth = depth[feature_alignment.start1:feature_alignment.end1] #/ median_depth
if not feature_alignment.forward1: cds_depth = cds_depth[::-1]
cds_ambiguous_depth = ambiguous_depth[feature_alignment.start1:feature_alignment.end1] #/ median_ambiguous_depth
if not feature_alignment.forward1: cds_ambiguous_depth = cds_ambiguous_depth[::-1]
cds_depth_expect = depth_expect[feature_alignment.start1:feature_alignment.end1]
if not feature_alignment.forward1: cds_depth_expect = cds_depth_expect[::-1]
#cds_average_depth_ratio = numpy.average(depth[feature_alignment.start1:feature_alignment.end1]) / median_depth
#cds_average_ambiguous_depth_ratio = numpy.average(ambiguous_depth[feature_alignment.start1:feature_alignment.end1]) / median_ambiguous_depth
#line += '%.1f\t' % cds_average_depth_ratio
#line += '%.1f\t' % cds_average_ambiguous_depth_ratio
#line += '%.1f..%.1f\t' % (numpy.minimum.reduce(cds_depth)/median_depth, numpy.maximum.reduce(cds_depth)/median_depth)
#line += '%.1f+/-%.1f\t' % (numpy.average(cds_depth)/median_depth, numpy.var(cds_depth)**0.5/median_depth)
#line += '%.1f..%.1f\t' % (numpy.minimum.reduce(cds_ambiguous_depth)/median_ambiguous_depth, numpy.maximum.reduce(cds_ambiguous_depth)/median_ambiguous_depth)
avg_expect = numpy.average(cds_depth_expect)
if avg_expect > 0.0:
cds_avg_depth = numpy.average(cds_depth)/avg_expect
cds_avg_ambiguous_depth = numpy.average(cds_ambiguous_depth)/avg_expect/ambiguous_factor
strange = (
(cds_depth >= cds_depth_expect*1.5) |
(cds_ambiguous_depth <= cds_depth_expect*(0.5*ambiguous_factor))
)
interesting_coverage = numpy.average(strange) >= coverage_cutoff
if interesting_coverage or diffs or diff_start or shifts or len(new_protein) != len(protein):
line = name + '\t' + locus_tag + '\t' + \
'%d\t' % (len(protein)-1) + \
'%d\t' % (len(new_protein)-1) + \
'%d\t' % len(diffs)
if use_coverage:
if avg_expect <= 0.0:
line += '\t\t\t'
else:
line += '%.1f\t' % (cds_avg_depth) + graphlet(cds_depth, cds_depth_expect)+'\t'
line += '%.1f\t' % (cds_avg_ambiguous_depth) + graphlet(cds_ambiguous_depth, cds_depth_expect*ambiguous_factor)+'\t'
line += '%.1f%%\t' % (numpy.average(cds_ambiguous_depth > 0.0)*100.0)
line += '%s\t' % feature.qualifiers.get('gene',[''])[0] + \
'%s' % feature.qualifiers.get('product',[''])[0]
notes = [ ]
if use_coverage and 'X' in new_protein:
xs = new_protein.count('X')
if xs == len(new_protein)-1: #First is M, so len-1
notes.append('\ No consensus')
else:
notes.append('\ No consensus for %d aa' % (new_protein.count('X')))
if len(new_protein) < len(protein):
notes.append('\ Shorter by %d aa' % (len(protein)-len(new_protein)))
if len(new_protein) > len(protein):
notes.append('\ Longer by %d aa' % (len(new_protein)-len(protein)))
if diff_start:
notes.append('\ Start changed: %s -> %s' % (dna[:3], new_dna[:3]))
if new_dna[:3] not in start_codons:
notes.append(' No longer a start codon!')
if shifts:
notes.append('\ Indels:')
for pos, old, new in shifts:
notes.append(' base %5d / codon %5d %s -> %s' % (pos+1,(pos//3)+1,old,new or '-'))
if diffs:
if verbose:
notes.append('\ Amino acid changes:')
for i, j, k in diffs:
notes.append(' codon %5d %s->%s (%s->%s)' % (
j+1,
protein_ali[i],
new_protein_ali[i],
dna[j*3:j*3+3] if protein_ali[i] != '-' else '-',
new_dna[k*3:k*3+3] if new_protein_ali[i] != '-' else '-'
))
#if len(new_protein) > len(protein):
# print 'New protein is longer:', new_protein[len(protein):]
#if len(new_protein) < len(protein):
# print 'New protein is shorter:', protein[len(new_protein):]
#print protein
#print new_protein
if tabular:
print line + '\t' + ' '.join([ ' '.join(note.strip().split()) for note in notes ])
else:
print line
for note in notes:
print '\t' + note
return 0
def pairs_command(args):
grace.expect_no_further_options(args)
assert len(args) == 2, 'Expected exactly two files in "pairs"'
reads_filenames.append(
[os.path.abspath(filename) for filename in args])
def main(args):
title1, args = grace.get_option_value(args, '--title1', str, None)
title2, args = grace.get_option_value(args, '--title2', str, None)
grace.expect_no_further_options(args)
if len(args) != 3:
print >> sys.stderr, USAGE
return 1
working_dir1 = args[0]
working_dir2 = args[1]
cutoff = float(args[2])
sequence_names = [
name for name, sequence in io.read_sequences(
os.path.join(working_dir1, 'reference.fa'))
]
if title1 is None:
title1 = working_dir1
if title2 is None:
title2 = working_dir2
n = 1
while significance([('A', n)], [('T', n)], 1.0) > cutoff:
n += 1
print '%g\tsignificance cutoff' % cutoff
print '%d\tdepth required to call substitution (greater if there are errors in the reads)' % n
print 'Sequence\tPosition in reference\tChange type\tReference\t%s\t%s\tp-value (no correction for multiple testing)\t%s\t%s' % (
title1, title2, title1, title2)
for sequence_name in sequence_names:
filename1 = os.path.join(
working_dir1,
grace.filesystem_friendly_name(sequence_name) + '-evidence.txt')
filename2 = os.path.join(
working_dir2,
grace.filesystem_friendly_name(sequence_name) + '-evidence.txt')
for (pos1, ins1, sub1, ref1, conins1,
consub1), (pos2, ins2, sub2, ref2, conins2,
consub2) in itertools.izip(read_file(filename1),
read_file(filename2)):
assert pos1 == pos2 and ref1 == ref2
if pos1 % 1000 == 0:
grace.status('Testing %s %d' % (sequence_name, pos1))
dec_ins1 = io.decode_evidence(ins1)
dec_ins2 = io.decode_evidence(ins2)
if dec_ins1 and dec_ins2:
sig = significance(io.decode_evidence(ins1),
io.decode_evidence(ins2), cutoff)
if sig is not None and sig <= cutoff:
grace.status('')
print '%s\t%d\t%s\t\t%s\t%s\t%g\t%s\t%s' % (
sequence_name, pos1, 'insertion-before', ins1, ins2,
sig, conins1, conins2)
dec_sub1 = io.decode_evidence(sub1)
dec_sub2 = io.decode_evidence(sub2)
if dec_sub1 and dec_sub2:
sig = significance(dec_sub1, dec_sub2, cutoff)
if sig is not None and sig <= cutoff:
if dec_sub1[0][0] == '-' or dec_sub2[0][0] == '-':
what = 'deletion'
elif dec_sub1[0][0] != dec_sub2[0][0]:
what = 'substitution'
else:
what = 'different mix'
grace.status('')
print '%s\t%d\t%s\t%s\t%s\t%s\t%g\t%s\t%s' % (
sequence_name, pos1, what, ref1, sub1, sub2, sig,
consub1, consub2)
grace.status('')
return 0
def reads_command(args):
grace.expect_no_further_options(args)
reads_filenames.extend([ [ os.path.abspath(filename) ] for filename in args])
def main(args):
title1, args = grace.get_option_value(args, "--title1", str, None)
title2, args = grace.get_option_value(args, "--title2", str, None)
grace.expect_no_further_options(args)
if len(args) != 3:
print >> sys.stderr, USAGE
return 1
working_dir1 = args[0]
working_dir2 = args[1]
cutoff = float(args[2])
sequence_names = [name for name, sequence in io.read_sequences(os.path.join(working_dir1, "reference.fa"))]
if title1 is None:
title1 = working_dir1
if title2 is None:
title2 = working_dir2
n = 1
while significance([("A", n)], [("T", n)], 1.0) > cutoff:
n += 1
print "%g\tsignificance cutoff" % cutoff
print "%d\tdepth required to call substitution (greater if there are errors in the reads)" % n
print "Sequence\tPosition in reference\tChange type\tReference\t%s\t%s\tp-value (no correction for multiple testing)\t%s\t%s" % (
title1,
title2,
title1,
title2,
)
for sequence_name in sequence_names:
filename1 = os.path.join(working_dir1, grace.filesystem_friendly_name(sequence_name) + "-evidence.txt")
filename2 = os.path.join(working_dir2, grace.filesystem_friendly_name(sequence_name) + "-evidence.txt")
for (pos1, ins1, sub1, ref1, conins1, consub1), (pos2, ins2, sub2, ref2, conins2, consub2) in itertools.izip(
read_file(filename1), read_file(filename2)
):
assert pos1 == pos2 and ref1 == ref2
if pos1 % 1000 == 0:
grace.status("Testing %s %d" % (sequence_name, pos1))
dec_ins1 = io.decode_evidence(ins1)
dec_ins2 = io.decode_evidence(ins2)
if dec_ins1 and dec_ins2:
sig = significance(io.decode_evidence(ins1), io.decode_evidence(ins2), cutoff)
if sig is not None and sig <= cutoff:
grace.status("")
print "%s\t%d\t%s\t\t%s\t%s\t%g\t%s\t%s" % (
sequence_name,
pos1,
"insertion-before",
ins1,
ins2,
sig,
conins1,
conins2,
)
dec_sub1 = io.decode_evidence(sub1)
dec_sub2 = io.decode_evidence(sub2)
if dec_sub1 and dec_sub2:
sig = significance(dec_sub1, dec_sub2, cutoff)
if sig is not None and sig <= cutoff:
if dec_sub1[0][0] == "-" or dec_sub2[0][0] == "-":
what = "deletion"
elif dec_sub1[0][0] != dec_sub2[0][0]:
what = "substitution"
else:
what = "different mix"
grace.status("")
print "%s\t%d\t%s\t%s\t%s\t%s\t%g\t%s\t%s" % (
sequence_name,
pos1,
what,
ref1,
sub1,
sub2,
sig,
consub1,
consub2,
)
grace.status("")
return 0
def main(args):
genbank_filename, args = grace.get_option_value(args,'--gbk',str,None)
use_indels, args = grace.get_option_value(args,'--indels',grace.as_bool,True)
use_reference, args = grace.get_option_value(args,'--reference',grace.as_bool,True)
give_evidence, args = grace.get_option_value(args,'--evidence',grace.as_bool,True)
give_consequences, args = grace.get_option_value(args,'--consequences',grace.as_bool,True)
require_all, args = grace.get_option_value(args,'--require-all',grace.as_bool,False)
require_bisect, args = grace.get_option_value(args,'--require-bisect',grace.as_bool,False)
full_output, args = grace.get_option_value(args,'--full',grace.as_bool,False)
format, args = grace.get_option_value(args,'--as',str,'table')
# Secret option!
limit, args = grace.get_option_value(args,'--limit',int,None)
grace.expect_no_further_options(args)
if len(args) < 1:
sys.stderr.write(USAGE)
return 1
working_dirs = [ ]
split_a = [ ]
split_b = [ ]
def default(args):
working_dirs.extend(args)
def splitting(args):
split_a.extend(args)
def splitting_from(args):
split_b.extend(args)
grace.execute(args, {
'splitting' : splitting,
'from' : splitting_from
}, default
)
if use_reference:
names = ['reference']
evidence_start = 1
else:
names = [ ]
evidence_start = 0
names.extend( norm_name(item) for item in working_dirs )
references = io.read_sequences(os.path.join(working_dirs[0], 'reference.fa'))
annotations = { }
if genbank_filename:
from Bio import SeqIO
for record in SeqIO.parse(io.open_possibly_compressed_file(genbank_filename),'genbank'):
sequence = record.seq.tostring()
features = [ item for item in record.features if item.type != 'source' ]
features.sort(key=lambda item: item.location.nofuzzy_start)
annotations[sequence] = features
iterator = reader(working_dirs, references, use_reference, annotations)
if not use_indels:
iterator = itertools.ifilter(has_no_indels, iterator)
if require_all or require_bisect or format == 'counts':
iterator = itertools.ifilter(fully_unambiguous, iterator)
if require_bisect:
iterator = itertools.ifilter(is_binary_partition, iterator)
if not require_bisect:
if full_output:
iterator = itertools.ifilter(not_boring_insertion, iterator)
else:
iterator = itertools.ifilter(is_interesting, iterator)
if split_a or split_b:
assert len(names) == len(set(names)), 'Two samples with the same name'
try:
split_a = [ names.index(norm_name(item)) for item in split_a ]
split_b = [ names.index(norm_name(item)) for item in split_b ]
except ValueError:
raise grace.Error('Sample to be split is not amongst samples given')
iterator = itertools.ifilter(is_split(split_a, split_b), iterator)
if limit:
iterator = itertools.islice(iterator, limit)
if format == 'table':
line = 'Reference\tPosition\tChange type'
line += '\t' + '\t'.join(names)
if give_evidence:
line += '\t' + '\t'.join(names[evidence_start:])
if give_consequences:
line += '\t' + '\t'.join(names[evidence_start:])
if annotations:
line += '\tAnnotations'
print line
for calls in iterator:
line = '%s\t%d\t%s\t%s' % (
calls.ref_name,
calls.ref_pos+1,
change_type(calls),
'\t'.join(item.consensus for item in calls.calls))
if give_evidence:
line += '\t' + '\t'.join(item.evidence for item in calls.calls[evidence_start:])
if give_consequences:
line += '\t' + '\t'.join(item.consequences for item in calls.calls[evidence_start:])
if annotations:
line += '\t' + describe_features(calls.features)
print line
elif format == 'compact':
for line in transpose_strings(names):
print line
print
for calls in iterator:
if calls.is_insertion:
footer = '%12d.5 %s' % (calls.ref_pos, calls.ref_name)
else:
footer = '%12d %s' % (calls.ref_pos+1, calls.ref_name)
t = transpose_strings([ item.consensus for item in calls.calls ], '-', 1)
top = t[0] + ' ' + footer
if give_consequences:
consequences = [ ]
for call in calls.calls:
if call.consequences:
for item in call.consequences.split(', '):
item = ' '.join(item.split()[:3])
if item not in consequences: consequences.append(item)
if consequences:
top += ' ' + ' / '.join(sorted(consequences))
top += ' ' + describe_features(calls.features)
print top
for line in t[1:]:
print line
elif format == 'nexus':
buckets = [ [ ] for name in names ]
for calls in iterator:
for i, char in enumerate(partition_string(calls)):
buckets[i].append(char)
print '#NEXUS'
print 'begin taxa;'
print 'dimensions ntax=%d;' % len(names)
print 'taxlabels'
for name in names:
print name
print ';'
print 'end;'
print 'begin characters;'
print 'dimensions nchar=%d;' % len(buckets[0])
print 'format datatype=STANDARD symbols="ACGT-0123456789" missing=N;'
print 'matrix'
for name, bucket in itertools.izip(names, buckets):
print name, ''.join(bucket)
print ';'
print 'end;'
elif format == 'counts':
for line in transpose_strings(names):
print line
print
counts = { }
for calls in iterator:
count_str = partition_string(calls)
if count_str not in counts:
counts[count_str] = 1
else:
counts[count_str] += 1
for count_str in sorted(counts, key=lambda x: (counts[x], x), reverse=True):
print '%s %d' % (transpose_strings(count_str)[0], counts[count_str])
else:
raise grace.Error('Unknown output format: ' + format)
def reads_command(args):
grace.expect_no_further_options(args)
reads_filenames.extend([[os.path.abspath(filename)]
for filename in args])
def old_main(args):
use_indels, args = grace.get_option_value(args,'--indels',int,1)
use_reference, args = grace.get_option_value(args,'--reference',int,1)
make_list, args = grace.get_option_value(args,'--list',int,0)
fasta_output, args = grace.get_option_value(args,'--fasta',int,0)
grace.expect_no_further_options(args)
if len(args) < 1:
sys.stderr.write(USAGE)
return 1
if fasta_output and use_indels:
print >> sys.stderr, 'Indels will not be included in FASTA output'
use_indels = 0
working_dirs = args
#reference_data = { } # (ref_name, position, change_type) -> string
#strain_data = { } # working_dir -> (ref_name, position, change_type) -> string
names = ['reference'] + working_dirs
substitution_calls = { } # ref_name -> [ [ call ] ]
insertion_calls = { } # ref_name -> [ [ call ] ]
substitution_evidence = { }
insertion_evidence = { }
for name, sequence in io.read_sequences(os.path.join(working_dirs[0], 'reference.fa')):
substitution_calls[name] = [ list(sequence.upper()) ]
insertion_calls[name] = [ [ '-' ] * len(sequence) ]
substitution_evidence[name] = [ [ '' ] * len(sequence) ]
insertion_evidence[name] = [ [ '' ] * len(sequence) ]
for working_dir in working_dirs:
for name in substitution_calls:
filename = os.path.join(working_dir, grace.filesystem_friendly_name(name) + '-evidence.txt')
f = open(filename,'rb')
this_substitution_calls = [ ]
this_insertion_calls = [ ]
this_substitution_evidence = [ ]
this_insertion_evidence = [ ]
header = f.readline()
if header.count('\t') != 5:
print >> sys.stderr, 'Old style evidence file. Please re-run nesoni consensus.'
return 1
for line in f:
fields = line.rstrip('\n').split('\t')
this_substitution_calls.append(fields[5])
this_insertion_calls.append(fields[4])
this_substitution_evidence.append(fields[2])
this_insertion_evidence.append(fields[1])
substitution_calls[name].append(this_substitution_calls)
insertion_calls[name].append(this_insertion_calls)
substitution_evidence[name].append(this_substitution_evidence)
insertion_evidence[name].append(this_insertion_evidence)
if not use_reference:
names.pop(0)
for name in substitution_calls:
substitution_calls[name].pop(0)
insertion_calls[name].pop(0)
substitution_evidence[name].pop(0)
insertion_evidence[name].pop(0)
interesting = find_interesting('substitution', substitution_calls, substitution_evidence)
if use_indels:
interesting.extend( find_interesting('insertion-before', insertion_calls, insertion_evidence) )
if not use_indels:
interesting = [ item for item in interesting if '-' not in item[3] ]
interesting.sort()
if fasta_output:
do_fasta_output(names, interesting)
return 0
#strain_reference_having_consensus = { } # working_dir -> ref_name -> string
#
#for working_dir in working_dirs:
# assert working_dir not in strain_data, 'Working directory given twice'
# strain_data[working_dir] = { }
#
# report_file = open(os.path.join(working_dir, 'report.txt'), 'rU')
# report_file.readline()
# for line in report_file:
# ref_name, position, change_type, old, new, evidence = \
# line.rstrip('\n').split('\t')
#
# if change_type == 'deletion':
# change_type = 'substitution'
#
# if not use_indels and \
# (change_type == 'insertion-before' or new == '-'):
# continue
#
# key = (ref_name, int(position), change_type)
# if key in reference_data:
# assert reference_data[key] == old
# else:
# reference_data[key] = old
#
# strain_data[working_dir][key] = new
# report_file.close()
#
# strain_reference_having_consensus[working_dir] = { }
# ref_have_con_filename = os.path.join(working_dir, 'reference_having_consensus.fa')
# for name, sequence in io.read_fasta(ref_have_con_filename):
# strain_reference_having_consensus[working_dir][name] = sequence
#
#keys = sorted(reference_data)
#
##Fill in any blanks
#for working_dir in working_dirs:
# for key in keys:
# if key in strain_data[working_dir]: continue
#
# # - Positions in report files start from 1 not 0
# # - Insertions must be bracketed
# lacks_consensus = (
# strain_reference_having_consensus[working_dir][key[0]][key[1]-1] == 'N' or
# (key[2] == 'insertion-before' and key[1] > 1 and
# strain_reference_having_consensus[working_dir][key[0]][key[1]-2] == 'N')
# )
#
# #If there's no consensus, record it as ambiguous
# if lacks_consensus:
# strain_data[working_dir][key] = 'N'
# else:
# strain_data[working_dir][key] = reference_data[key]
#all_data_names = ([ 'reference' ] if use_reference else []) + working_dirs
#all_data = ([ reference_data ] if use_reference else []) + \
# [ strain_data[working_dir] for working_dir in working_dirs ]
#all_data_names = ([ 'reference' ] if use_reference else []) + working_dirs
ones = ( 1 << len(names) )-1
total_differences = 0
if make_list:
print '\t'.join(['Partition','Sequence','Position in reference','Change type'] + names + names)
for i in xrange(1,(1<<len(names))-1,2):
set1 = [ ]
set2 = [ ]
for j in xrange(len(names)):
if i & (1<<j):
set1.append(j)
else:
set2.append(j)
if make_list:
print
print ', '.join( names[i] for i in set1 ) + ' vs ' + \
', '.join( names[i] for i in set2 )
print
n = 0
for refname, position, change_type, values, has_ambiguous, evidence in interesting:
#Skip if *any* ambiguity
if has_ambiguous:
continue
if any( values[i] != values[set1[0]] for i in set1[1:] ) or \
any( values[i] != values[set2[0]] for i in set2[1:] ):
continue
if make_list:
if change_type == 'substitution' and '-' in values: change_type = 'deletion'
print '\t%s\t%d\t%s\t' % (refname,position,change_type) + '\t'.join(values) + '\t' + '\t'.join(evidence)
n += 1
total_differences += n
if not make_list:
print ', '.join( names[i] for i in set1 ) + ' vs ' + \
', '.join( names[i] for i in set2 ) + \
': %d differences' %n
if not make_list:
print
print 'Total: %d' % total_differences
if make_list:
print
print 'Ignored'
print
n_multiway = 0
n_ambiguous = 0
for refname, position, change_type, values, has_ambiguous, evidence in interesting:
confusing = False
if has_ambiguous:
n_ambiguous += 1
confusing = True
elif len(set(values)) > 2:
n_multiway += 1
confusing = True
if make_list and confusing:
print '\t%s\t%d\t%s\t' % (refname,position,change_type) + '\t'.join(values) + '\t' + '\t'.join(evidence)
if not make_list:
print
print 'Ambiguities ignored: %d' % n_ambiguous
print 'Multi-way changes ignored: %d' % n_multiway
assert total_differences + n_ambiguous + n_multiway == len(interesting)
return 0
def recombination(args):
grace.expect_no_further_options(args)
if len(args) != 2:
print >> sys.stderr, USAGE
raise grace.Help_shown()
working_dir, seq_name = args
references = dict(io.read_sequences(os.path.join(working_dir, 'reference.fa')))
depth = { }
prefixes = { }
suffixes = { }
for name in references:
depth[name] = numpy.zeros(len(references[name]), 'int64')
prefixes[name] = [ [] for base in references[name] ]
suffixes[name] = [ [] for base in references[name] ]
def register_divergence(hit):
if not hit.query_forward:
hit = hit.reversed()
margin = 20
if hit.target_end - hit.target_start < 20:
return False
depth[hit.target_name][hit.target_start : hit.target_end] += 1
any = False
if hit.query_end <= len(hit.query_seq)-margin: # and hit.target_end < len(hit.target_seq):
suffixes[hit.target_name][hit.target_end-1].append( hit.query_seq[hit.query_end:] )
any = True
if hit.query_start >= margin: # and hit.target_start > 0:
prefixes[hit.target_name][hit.target_start].append( hit.query_seq[:hit.query_start] )
any = True
return any
n = 0
for (read_name, read_seq), hits in shrimp.iter_read_hits(working_dir):
# Skip reads containing Ns
if 'N' in read_seq: continue
for line in hits:
register_divergence(alignment_from_shrimp(line, references, read_name, read_seq))
n += 1
#if n > 100000:
# break
if n%10000 == 0:
grace.status('Processing read %s' % grace.pretty_number(n))
grace.status('')
def show_items(items):
original_length = len(items)
cut = 0
while len(items) > 80:
cut += 1
items = [ item for item in items if item[0] >= cut ]
for item in items:
print item[1]
if len(items) < original_length:
print '(and %d more occurring %d times or less)' % (original_length-len(items), cut-1)
def score(items):
if not items: return 1.0
return float(sum( item[0] * item[0] for item in items )) / (sum( item[0] for item in items )**2)
def summarize_prefixes(seqs, pad):
seqs = sorted(seqs, key=lambda seq: seq[::-1])
cut = 100
while True:
items = [ ]
for (seq, iterator) in itertools.groupby(seqs, key = lambda x: x[-cut:]):
ss = list(iterator)
anylong = any( item != seq for item in ss )
n = len(ss)
items.append( (n, ('%'+str(pad)+'s')%(('...' if anylong else '') + seq) + ' x %d' % n) )
if score(items) >= 1.0/20: break
cut -= 1
show_items(items)
def summarize_suffixes(seqs, pad):
seqs = sorted(seqs)
cut = 100
while True:
items = [ ]
for (seq, iterator) in itertools.groupby(seqs, key = lambda x: x[:cut]):
ss = list(iterator)
anylong = any( item != seq for item in ss )
n = len(ss)
items.append( (n, ('%'+str(pad)+'s')%('%d x '%n) + seq + ('...' if anylong else '')) )
if score(items) >= 1.0/20: break
cut -= 1
show_items(items)
print 'Position Depth Changed prefixes Changed suffixes'
print ' Count % of depth Count % of depth'
for i in xrange(len(references[seq_name])):
print '%8d %10d %9d %11s %9d %11s' % (
i+1,
depth[seq_name][i],
len(prefixes[seq_name][i]),
'%.3f%%' % (len(prefixes[seq_name][i])*100.0/depth[seq_name][i]) if prefixes[seq_name][i] else '',
len(suffixes[seq_name][i]),
'%.3f%%' % (len(suffixes[seq_name][i])*100.0/depth[seq_name][i]) if suffixes[seq_name][i] else '')
#summarize_suffixes(suffixes[name][i], references[name][i+1:], references[name], suffix_depth[name][i])
print
print 'Details'
print
for i in xrange(len(references[seq_name])):
print '%-80s*' % ('Base %d' % (i+1))
print pad_slice(references[seq_name], i-80,i+1+80)
summarize_prefixes(prefixes[seq_name][i], 80)
summarize_suffixes(suffixes[seq_name][i], 81)
print
def default(args):
grace.expect_no_further_options(args)
if len(args) != 1:
print >> sys.stderr, BATCH_HELP % default_nesoni
raise grace.Help_shown()
options.dirname = args[0]
def reference(args):
grace.expect_no_further_options(args)
options.references.extend(args)
def reads(args):
grace.expect_no_further_options(args)
sample.reads.extend(args)
def main(args):
mincov, args = grace.get_option_value(args, '--mincov', int, 1)
maxdiff, args = grace.get_option_value(args, '--maxdiff', int, 16)
minsize, args = grace.get_option_value(args, '--minsize', int, 200)
what, args = grace.get_option_value(args, '--what', as_core_or_unique, 'core')
is_core = (what == 'core')
grace.expect_no_further_options(args)
if len(args) < 2:
print >> sys.stderr, HELP
raise grace.Help_shown()
output_dir, working_dirs = args[0], args[1:]
assert not path.exists(path.join(output_dir, 'reference.fa')), \
'Output directory not given'
if not path.exists(output_dir):
os.mkdir(output_dir)
for name, seq in io.read_sequences(path.join(working_dirs[0],'reference.fa')):
print name
friendly_name = grace.filesystem_friendly_name(name)
good = [ True ] * len(seq)
for working_dir in working_dirs:
if is_core:
suffix = '-depth.userplot'
else:
suffix = '-ambiguous-depth.userplot'
data = trivia.read_unstranded_userplot(
os.path.join(working_dir, friendly_name+suffix)
)
assert len(seq) == len(data)
for i in xrange(len(seq)):
if good[i]:
if is_core:
good[i] = data[i] >= mincov
else:
good[i] = data[i] < mincov
#Close holes
start = -maxdiff-1
n_holes = 0
for i in xrange(len(seq)):
if good[i]:
if 0 < i-start <= maxdiff:
for j in xrange(start,i): good[j] = True
n_holes += 1
start = i+1
print 'Closed', grace.pretty_number(n_holes), 'holes'
f = open(path.join(output_dir, '%s-%s.fa' % (friendly_name,what)), 'wb')
io.write_fasta(f, name,
''.join([ (seq[i] if good[i] else 'N')
for i in xrange(len(seq)) ])
)
f.close()
f = open(path.join(output_dir, '%s-%s_masked.fa' % (friendly_name,what)), 'wb')
io.write_fasta(f, name,
''.join([ (seq[i] if good[i] else seq[i].lower())
for i in xrange(len(seq)) ])
)
f.close()
f_good = open(path.join(output_dir, '%s-%s_parts.fa' % (friendly_name,what)), 'wb')
f_nongood = open(path.join(output_dir, '%s-non%s_parts.fa' % (friendly_name,what)), 'wb')
start = 0
n_good = [0]
n_good_bases = [0]
def emit(i):
if i-start < minsize: return
if good[start]:
n_good[0] += 1
n_good_bases[0] += i-start
io.write_fasta(
f_good if good[start] else f_nongood,
'%s:%d..%d' % (name, start+1,i),
seq[start:i]
)
for i in xrange(1,len(seq)):
if good[i] != good[start]:
emit(i)
start = i
emit(len(seq))
f_nongood.close()
f_good.close()
print grace.pretty_number(sum(good)), 'bases are '+what+', of', grace.pretty_number(len(seq)), 'in reference sequence'
print grace.pretty_number(n_good[0]), 'parts at least', grace.pretty_number(minsize), 'bases long with', grace.pretty_number(n_good_bases[0]), 'total bases'
print
def main(args):
mincov, args = grace.get_option_value(args, '--mincov', int, 1)
maxdiff, args = grace.get_option_value(args, '--maxdiff', int, 16)
minsize, args = grace.get_option_value(args, '--minsize', int, 200)
what, args = grace.get_option_value(args, '--what', as_core_or_unique,
'core')
is_core = (what == 'core')
grace.expect_no_further_options(args)
if len(args) < 2:
print >> sys.stderr, HELP
raise grace.Help_shown()
output_dir, working_dirs = args[0], args[1:]
assert not path.exists(path.join(output_dir, 'reference.fa')), \
'Output directory not given'
if not path.exists(output_dir):
os.mkdir(output_dir)
for name, seq in io.read_sequences(
path.join(working_dirs[0], 'reference.fa')):
print name
friendly_name = grace.filesystem_friendly_name(name)
good = [True] * len(seq)
for working_dir in working_dirs:
if is_core:
suffix = '-depth.userplot'
else:
suffix = '-ambiguous-depth.userplot'
data = trivia.read_unstranded_userplot(
os.path.join(working_dir, friendly_name + suffix))
assert len(seq) == len(data)
for i in xrange(len(seq)):
if good[i]:
if is_core:
good[i] = data[i] >= mincov
else:
good[i] = data[i] < mincov
#Close holes
start = -maxdiff - 1
n_holes = 0
for i in xrange(len(seq)):
if good[i]:
if 0 < i - start <= maxdiff:
for j in xrange(start, i):
good[j] = True
n_holes += 1
start = i + 1
print 'Closed', grace.pretty_number(n_holes), 'holes'
f = open(path.join(output_dir, '%s-%s.fa' % (friendly_name, what)),
'wb')
io.write_fasta(
f, name,
''.join([(seq[i] if good[i] else 'N') for i in xrange(len(seq))]))
f.close()
f = open(
path.join(output_dir, '%s-%s_masked.fa' % (friendly_name, what)),
'wb')
io.write_fasta(
f, name, ''.join([(seq[i] if good[i] else seq[i].lower())
for i in xrange(len(seq))]))
f.close()
f_good = open(
path.join(output_dir, '%s-%s_parts.fa' % (friendly_name, what)),
'wb')
f_nongood = open(
path.join(output_dir, '%s-non%s_parts.fa' % (friendly_name, what)),
'wb')
start = 0
n_good = [0]
n_good_bases = [0]
def emit(i):
if i - start < minsize: return
if good[start]:
n_good[0] += 1
n_good_bases[0] += i - start
io.write_fasta(f_good if good[start] else f_nongood,
'%s:%d..%d' % (name, start + 1, i), seq[start:i])
for i in xrange(1, len(seq)):
if good[i] != good[start]:
emit(i)
start = i
emit(len(seq))
f_nongood.close()
f_good.close()
print grace.pretty_number(
sum(good)), 'bases are ' + what + ', of', grace.pretty_number(
len(seq)), 'in reference sequence'
print grace.pretty_number(
n_good[0]), 'parts at least', grace.pretty_number(
minsize), 'bases long with', grace.pretty_number(
n_good_bases[0]), 'total bases'
print
