def __init__(self, fhand, reference_name, grouping=None):
'It inits the class'
# The fhand is as it arrives
open(fhand.name, 'w')
self.fhand = open(fhand.name, 'a')
self._namer = SnvNamer()
self._temp_fhand = NamedTemporaryFile(mode='a')
self._filter_descriptions = {}
self._header = []
if grouping is None:
grouping = 'read_groups'
self._genotype_grouping_key = grouping
self._genotype_groups = OrderedDict()
self._get_pre_header(reference_name)
self.num_features = 0
class VariantCallFormatWriter(object):
'It writes variant call format files for the snvs.'
def __init__(self, fhand, reference_name, grouping=None):
'It inits the class'
# The fhand is as it arrives
open(fhand.name, 'w')
self.fhand = open(fhand.name, 'a')
self._namer = SnvNamer()
self._temp_fhand = NamedTemporaryFile(mode='a')
self._filter_descriptions = {}
self._header = []
if grouping is None:
grouping = 'read_groups'
self._genotype_grouping_key = grouping
self._genotype_groups = OrderedDict()
self._get_pre_header(reference_name)
self.num_features = 0
def _get_pre_header(self, reference_name):
'It writes the header of the vcf file'
header = self._header
header.append('##fileformat=VCFv4.1')
header.append('##fileDate=%s' %
datetime.date.today().strftime('%Y%m%d'))
header.append('##source=franklin')
header.append('##reference=%s' % reference_name)
header.append('##INFO=<ID=NS,Number=1,Type=Integer,Description="Number of Samples With Data">')
header.append('##INFO=<ID=AF,Number=A,Type=Float,Description="Allele Frequency">')
header.append('##INFO=<ID=RC,Number=A,Type=Integer,Description="Read count of the alt alleles">')
header.append('##INFO=<ID=MQ,Number=1,Type=Float,Description="RMS Mapping Quality">')
header.append('##INFO=<ID=BQ,Number=1,Type=Float,Description="RMS Base Quality">')
header.append('##INFO=<ID=GC,Number=.,Type=String,Description="Genotype Counts: Num. genotypes in which every alleles has been detected">')
header.append('##INFO=<ID=GP,Number=1,Type=String,Description="Genotype polimorphism">')
header.append('##INFO=<ID=EZ,Number=1,Type=String,Description="CAP enzymes">')
header.append('##FORMAT=<ID=GT,Number=1,Type=String,Description="Read group Genotype">')
header.append('##FORMAT=<ID=EC,Number=.,Type=Integer,Description="Allele count for the ref and alt alleles in the order listed">')
def close(self):
'It merges the header and the snv data'
# Append the data spec to the header
fhand = self.fhand
self._add_filters_to_header()
line_items = ['#CHROM', 'POS', 'ID', 'REF', 'ALT', 'QUAL', 'FILTER',
'INFO', 'FORMAT']
line_items.extend([group.upper() for group in self._genotype_groups.keys()])
num_items_per_line = len(line_items)
self._header.append('%s\n' % '\t'.join(line_items))
fhand.write('\n'.join(self._header))
for line in open(self._temp_fhand.name):
#fix the missing genotype groups
line = line.strip()
line = line.split()
num_items = len(line)
line.extend(['.:.'] * (num_items_per_line - num_items))
line = '\t'.join(line)
#\fix the missing genotype groups
fhand.write(line + '\n')
fhand.flush()
def _add_filters_to_header(self):
'It adds the used filter tag to the header'
for name, desc in self._filter_descriptions.values():
filter_desc = '##FILTER=<ID=%s,Description="%s">' % (name, desc)
self._header.append(filter_desc)
def write(self, sequence):
'It writes the snvs present in the given sequence as SeqFeatures'
for snv in sequence.get_features(kind='snv'):
self.num_features += 1
self._write_snv(sequence, snv)
@staticmethod
def _create_alternative_alleles(alleles):
'It returns the ALT part on the vcf'
str_alleles = []
alternative_alleles = []
for allele in alleles:
kind = allele[1]
if kind == INVARIANT:
continue
str_allele = allele[0].replace('-', '')
str_alleles.append(str_allele)
alternative_alleles.append(allele)
if str_alleles:
str_alleles = ','.join(str_alleles)
else:
str_alleles = '.'
return str_alleles, alternative_alleles
def _create_filters(self, qualifiers):
'It returns the FILTER part on the vcf'
filter_strs = []
if 'filters' not in qualifiers:
return '.'
for name, filters_data in qualifiers['filters'].items():
for parameters, result in filters_data.items():
if not result:
continue
short_name, description = self._namer.get_filter_description(
name, parameters,
self._filter_descriptions)
short_name = short_name.upper()
filter_strs.append(short_name)
self._filter_descriptions[name, parameters] = (short_name,
description)
if not filter_strs:
return 'PASS'
else:
return ';'.join(filter_strs)
def _create_info(self, qualifiers, alternative_alleles):
'It creates the INFO bit on the vcf'
toprint_items = []
alleles = qualifiers['alleles']
#RC allele count in genotypes, for each ALT allele, in the same order as
#listed
acounts = [] #allele_count
for allele in alternative_alleles:
acount = _allele_count(allele, alleles, group_kind='read_groups')
acounts.append(acount)
if acounts:
toprint_items.append('RC=%s' % ','.join(map(str, acounts)))
#AF allele frequency for each ALT allele in the same order as listed:
reference_allele = qualifiers['reference_allele'], INVARIANT
if reference_allele in alleles:
ref_count = _allele_count(reference_allele, alleles,
group_kind='read_groups')
else:
ref_count = 0
total_count = float(sum(acounts) + ref_count)
afreqs = [acount / total_count for acount in acounts]
if afreqs:
toprint_items.append('AF=%s' % ','.join(map(lambda x: '%.1f' % x,
afreqs)))
#MQ RMS mapping quality, e.g. MQ=52
#BQ RMS base quality at this position
for kind, strfmt in (('mapping_quality', 'MQ=%.2f'),
('quality', 'BQ=%.2f')):
qual = qualifiers[kind]
if qual is not None:
toprint_items.append(strfmt % qual)
#genotype count
#we count in how many genotypes every allele has been found.
allele_counts = self._allele_count_by_group(alleles=alleles,
reference_allele=reference_allele[0],
alternative_alleles=alternative_alleles,
read_groups=qualifiers['read_groups'],
count_reads=False)
#if some allele is missing there are 0 counts of it
n_als = max(allele_counts.keys()) + 1
genotype_counts = [(al, len(allele_counts.get(al, [])))for al in range(n_als)]
#now we print
counts = [str(count[1]) for count in genotype_counts]
toprint_items.append('GC=%s' % (','.join(counts)))
#genotype polymorphism
#1 - (number_groups_for_the_allele_with_more_groups / number_groups)
number_of_groups = sum([count[1] for count in genotype_counts])
genotype_polymorphism = 1 - genotype_counts[0][1] / float(number_of_groups)
toprint_items.append('GP=%.2f' % genotype_polymorphism)
#cap enzymes
if 'cap_enzymes' in qualifiers and qualifiers['cap_enzymes']:
to_print = 'EZ=%s' % ','.join(qualifiers['cap_enzymes'])
toprint_items.append(to_print)
if toprint_items:
return ';'.join(toprint_items)
else:
return '.'
@staticmethod
def _create_quality(alleles, alternative_alleles):
'''It returns the quality for this snv
QUAL phred-scaled quality score for the assertion made in ALT. i.e. give
-10log_10 prob(call in ALT is wrong). If ALT is "." (no variant) then
this is -10log_10 p(variant), and if ALT is not "." this is -10log_10
p(no variant). High QUAL scores indicate high confidence calls.
Although traditionally people use integer phred scores, this field is
permitted to be a floating point so to enable higher resolution for low
confidence calls if desired. (Numeric, Missing Value: -1)'''
if alternative_alleles:
phreds = [alleles[allele]['quality'] for allele in alternative_alleles]
if len(phreds) == 1:
phred = phreds[0]
else:
inv_phred = lambda phred: math.pow(10, (-phred / 10))
probs = map(inv_phred, phreds[:2])
prob = probs[0] * probs[1]
phred = -10 * math.log10(prob)
else:
phred = alleles.values()[0]['quality']
return '%i' % phred
@staticmethod
def _numbers_for_alleles(reference_allele, alternative_alleles):
'It returns a key with the numbers for the alleles'
#a map from alleles to allele index (0 for reference, etc)
alleles_index = [(reference_allele, INVARIANT)]
alleles_index.extend(alternative_alleles)
alleles_index = dict(zip(alleles_index, range(len(alleles_index))))
return alleles_index
def _allele_count_by_group(self, alleles, reference_allele,
alternative_alleles, read_groups, count_reads):
'''It returns the allele counts by group
It can answer to two questions:
- How many times have the allele been found in every group?
(count_reads = True)
It returns a dict indexed by group and the alleles
(with the vcf number coding)
- in which groups the allele have been found?
(count_reads= False)
It returns a dict indexed by the alleles
(with the vcf number coding)
It takes into account the grouping_key (read_group, sample or library)
It requires the dict with the information about the read_groups.
'''
#a map from alleles to allele index (0 for reference, etc)
alleles_index = self._numbers_for_alleles(reference_allele,
alternative_alleles)
grouping_key = self._genotype_grouping_key
alleles_by_group = {}
for allele, allele_info in alleles.items():
#we need the index for the allele
# if allele not in alleles_index:
# continue
try:
allele_index = alleles_index[allele]
except KeyError:
print 'allele', allele
print "allele index", alleles_index
print 'ref_allele', reference_allele
print 'alternative_alleles', alternative_alleles
print 'alleles', alleles
raise
for read_group in allele_info['read_groups']:
group = _get_group(read_group, grouping_key, read_groups)
if group not in self._genotype_groups:
self._genotype_groups[group] = True
if count_reads:
if group not in alleles_by_group:
alleles_by_group[group] = {}
if allele_index not in alleles_by_group[group]:
alleles_by_group[group][allele_index] = 0
count = allele_info['read_groups'][read_group]
alleles_by_group[group][allele_index] += count
else:
if allele_index not in alleles_by_group:
alleles_by_group[allele_index] = set()
alleles_by_group[allele_index].add(group)
return alleles_by_group
def _create_genotypes(self, qualifiers, alternative_alleles):
'It returns the genotype section for this snv'
alleles = qualifiers['alleles']
reference_allele = qualifiers['reference_allele']
read_groups = qualifiers['read_groups']
items = []
#the format
items.append('GT:EC')
#a map from alleles to allele index (0 for reference, etc)
alleles_index = self._numbers_for_alleles(reference_allele,
alternative_alleles)
#now we need the alleles for every sample
alleles_by_group = self._allele_count_by_group(alleles=alleles,
reference_allele=reference_allele,
alternative_alleles=alternative_alleles,
read_groups=read_groups,
count_reads=True)
#now we can build the info for every sample
for group in self._genotype_groups.keys():
allele_counts = alleles_by_group.get(group, {None: None})
alleles, counts = [], []
for allele, count in allele_counts.items():
allele = str(allele) if allele is not None else '.'
alleles.append(str(allele))
count = str(count) if count is not None else '.'
counts.append(count)
#print group, alleles, counts
mix_genotype = '|'.join(alleles)
allele_counts = ','.join(counts)
items.append('%s:%s' % (mix_genotype, allele_counts))
#print 'result', ' '.join(items)
return '\t'.join(items)
def _write_snv(self, sequence, snv):
'Given an snv feature it writes a line in the vcf'
items = [] #items to write
items.append(get_seq_name(sequence))
items.append(str(int(snv.location.start.position) + 1))
id_ = snv.id
if id_ == "<unknown id>":
id_ = '.'
items.append(id_)
qualifiers = snv.qualifiers
ref_seq = qualifiers['reference_allele'].replace('-', '')
items.append(ref_seq)
toprint_af, alternative_alleles = self._create_alternative_alleles(
qualifiers['alleles'])
items.append(toprint_af)
items.append(self._create_quality(qualifiers['alleles'],
alternative_alleles))
filters = self._create_filters(qualifiers)
items.append(filters)
try:
items.append(self._create_info(qualifiers, alternative_alleles))
except KeyError:
print 'sequence', get_seq_name(sequence)
print 'position', str(int(snv.location.start.position))
raise
items.append(self._create_genotypes(qualifiers, alternative_alleles))
self._temp_fhand.write('%s\n' % '\t'.join(items))
self._temp_fhand.flush()
def test_get_filter_description():
'It tests get_filter_description function'
filter_name = 'close_to_intron'
parameters = 30
filter_descriptions = {}
namer = SnvNamer()
name, desc = namer.get_filter_description(filter_name, parameters,
filter_descriptions)
assert name == 'I30'
assert desc == 'An intron is located closer than 30 base pairs'
filter_name = 'maf'
parameters = 0.6
filter_descriptions = {}
name, desc = namer.get_filter_description(filter_name, parameters,
filter_descriptions)
assert name == 'maf1'
assert desc == 'The most frequent allele in All: All. frequency greater than 0.60'
filter_name = 'maf'
parameters = (0.6, ('1', '2'), 'read_group')
filter_descriptions = {}
name, desc = namer.get_filter_description(filter_name, parameters,
filter_descriptions)
assert name == 'maf2'
assert desc == 'The most frequent allele in read_group: 1,2. frequency greater than 0.60'
filter_name = 'by_kind'
parameters = SNP
filter_descriptions = {}
name, desc = namer.get_filter_description(filter_name, parameters,
filter_descriptions)
filter_name = 'is_variable'
kind = 'read_groups'
groups = ['rg1', 'rg2']
maf = 0.6
min_num_reads = 3
in_union = True
in_all_groups = True
reference_free = True
parameters = (kind, tuple(groups), in_union, in_all_groups,
reference_free, maf, min_num_reads)
filter_descriptions = {}
name, desc = namer.get_filter_description(filter_name, parameters,
filter_descriptions)
assert name[:3] == 'vrg'
descrip = "It is not variable, or no data, in the read_groups : rg1,rg2."
descrip += ' All together: True. maf:0.600000. min_num_reads:3'
assert desc == descrip
name, desc = namer.get_filter_description(filter_name, parameters,
filter_descriptions)
assert name[:3] == 'vrg'
filter_name = 'is_not_variable'
kind = 'read_groups'
groups = ['rg1', 'rg2']
maf = 0.7
min_num_reads = 2
in_union = True
in_all_groups = True
reference_free = True
parameters = (kind, tuple(groups), in_union, in_all_groups,
reference_free, maf, min_num_reads)
filter_descriptions = {}
name, desc = namer.get_filter_description(filter_name, parameters,
filter_descriptions)
assert name[:4] == 'nvrg'
descrip = "It is variable, or no data, in the read_groups : rg1,rg2."
descrip += ' All together: True. maf:0.700000. min_num_reads:2'
assert desc == descrip
filter_name = 'is_not_variable'
kind = 'read_groups'
groups = ['rg1', 'rg2']
maf = None
min_num_reads = None
in_union = True
in_all_groups = True
reference_free = True
parameters = (kind, tuple(groups), in_union, in_all_groups,
reference_free, maf, min_num_reads)
filter_descriptions = {}
name, desc = namer.get_filter_description(filter_name, parameters,
filter_descriptions)
assert name[:4] == 'nvrg'
descrip = "It is variable, or no data, in the read_groups : rg1,rg2."
descrip += ' All together: True'
assert desc == descrip
parameters = (4, 'read_groups')
name, desc = namer.get_filter_description('min_groups',
parameters,
filter_descriptions)
assert name == 'mr4'
assert desc == 'SNV read in less than 4 read_groups'
parameters = (True)
name, desc = namer.get_filter_description('cap_enzymes',
parameters,
filter_descriptions)
assert name == 'cet'
assert desc == 'SNV is not a CAP detectable by the enzymes: all'
parameters = (False)
name, desc = namer.get_filter_description('cap_enzymes',
parameters,
filter_descriptions)
assert name == 'cef'
assert desc == 'SNV is not a CAP detectable by the enzymes: cheap ones'