def testFromTabix(self):
# use ascii encoding - should raise error
t = pysam.TabixFile(self.tmpfilename + ".gz", encoding="ascii")
results = list(t.fetch(parser=pysam.asVCF()))
self.assertRaises(UnicodeDecodeError, getattr, results[1], "id")
t = pysam.TabixFile(self.tmpfilename + ".gz", encoding="utf-8")
results = list(t.fetch(parser=pysam.asVCF()))
self.assertEqual(getattr(results[1], "id"), u"Rene\xe9")
def testFromTabix(self):
# use ascii encoding - should raise error
t = pysam.TabixFile(self.tmpfilename + ".gz", encoding="ascii")
results = list(t.fetch(parser=pysam.asVCF()))
self.assertRaises(UnicodeDecodeError, getattr, results[1], "id")
t = pysam.TabixFile(self.tmpfilename + ".gz", encoding="utf-8")
results = list(t.fetch(parser=pysam.asVCF()))
self.assertEqual(getattr(results[1], "id"), u"Rene\xe9")
def fetch(self, chrom, start, end):
""" yield tuples (vcf object + info dict key->val) for a range
vcf row attributes are:
contig
pos chromosomal position, zero-based
id
ref reference
alt alt
qual qual
filter filter
info info
format format specifier.
"""
chrom = chrom.replace("chr", "")
#fname = self.fnameDict[chrom]
#vcf = pysam.VCF()
#vcf.connect(fname)
tbi = self.fhDict[chrom]
it = tbi.fetch(chrom, start, end, parser=pysam.asVCF())
for row in it:
infoDict = {}
infoStr = row.info
for keyVal in infoStr.split(";"):
if "=" not in keyVal:
continue
key, val = keyVal.split("=")
infoDict[key] = val
yield row, infoDict
def testRead(self):
ncolumns = len(self.columns)
for x, r in enumerate(self.tabix.fetch(parser=pysam.asVCF())):
c = self.compare[x]
for y, field in enumerate(self.columns):
# it is ok to have a missing format column
if y == 8 and y == len(c):
continue
if field == "pos":
self.assertEqual(int(c[y]) - 1, getattr(r, field))
self.assertEqual(int(c[y]) - 1, r.pos)
else:
self.assertEqual(
c[y], getattr(r, field),
"mismatch in field %s: %s != %s" %
(field, c[y], getattr(r, field)))
if len(c) == 8:
self.assertEqual(0, len(r))
else:
self.assertEqual(len(c), len(r) + ncolumns)
for y in range(len(c) - ncolumns):
self.assertEqual(c[ncolumns + y], r[y])
def __init__(self, file_path, parser=pysam.asVCF()):
self.vcf_file_path = file_path
self.tabix_file = pysam.TabixFile(file_path, parser=parser)
self.sample_names = self.read_sample_names()
self.clens = self.contig_lengths()
self.indexDelta = -1 if tuple(map(
int, pysam.__version__.split('.'))) > (0, 5, 0) else 0
def testRead(self):
ncolumns = len(self.columns)
for x, r in enumerate(self.tabix.fetch(parser=pysam.asVCF())):
c = self.compare[x]
for y, field in enumerate(self.columns):
# it is ok to have a missing format column
if y == 8 and y == len(c):
continue
if field == "pos":
self.assertEqual(int(c[y]) - 1, getattr(r, field))
self.assertEqual(int(c[y]) - 1, r.pos)
else:
self.assertEqual(c[y], getattr(r, field),
"mismatch in field %s: %s != %s" %
(field, c[y], getattr(r, field)))
if len(c) == 8:
self.assertEqual(0, len(r))
else:
self.assertEqual(len(c), len(r) + ncolumns)
for y in range(len(c) - ncolumns):
self.assertEqual(c[ncolumns + y], r[y])
self.assertEqual("\t".join(map(str, c)),
str(r))
def fetch(self, chrom, start, end):
""" yield tuples (vcf object + info dict key->val) for a range
vcf row attributes are:
contig
pos chromosomal position, zero-based
id
ref reference
alt alt
qual qual
filter filter
info info
format format specifier.
"""
chrom = chrom.replace("chr","")
#fname = self.fnameDict[chrom]
#vcf = pysam.VCF()
#vcf.connect(fname)
tbi = self.fhDict[chrom]
it = tbi.fetch(chrom, start, end, parser=pysam.asVCF())
for row in it:
infoDict = {}
infoStr = row.info
for keyVal in infoStr.split(";"):
if "=" not in keyVal:
continue
key, val = keyVal.split("=")
infoDict[key] = val
yield row, infoDict
class VcfReader:
"""
Read comfortably from VCF style files with main focus on chr, start, ref and alt fields.
Note that this API uses 1-based coordinates with both start and end included in the interval.
PySam API uses 0-based half-open intervals, so we have to convert internally.
"""
def __init__(self, input_file):
self.filename = input_file
self.indexed = False
if input_file.strip() == "-":
ifile = sys.stdin
elif input_file.endswith(".bz2"):
try:
ifile = bz2file.BZ2File(input_file, "r", buffering=0)
except Exception, e:
raise e
elif input_file.endswith(".gz") or input_file.endswith(".bgz"):
# try to open the file with Tabix
try:
ifile = pysam.Tabixfile(input_file, parser=pysam.asVCF())
self.indexed = True
except Exception, e:
try:
ifile = gzip.GzipFile(input_file, "r")
except Exception, e:
raise e
def filter_vcfs(genotype, contig, start, end):
if contig in genotype.contigs:
parser = pysam.asVCF()
# This raises a ValueError if the VCF does not
# contain any entries for the specified contig.
for vcf in genotype.fetch(contig, start, end, parser=parser):
if vcf.filter in ("PASS", "."):
yield vcf
def filter_vcfs(genotype, contig, start, end):
if contig in genotype.contigs:
parser = pysam.asVCF()
# This raises a ValueError if the VCF does not
# contain any entries for the specified contig.
for vcf in genotype.fetch(contig, start, end, parser=parser):
if vcf.filter in ("PASS", "."):
yield vcf
def fetch(self,
reference = None,
start = None,
end = None,
region = None ):
""" Parse a stream of VCF-formatted lines. Initializes class instance and return generator """
iter = self.tabixfile.fetch( reference, start, end, region, parser = pysam.asVCF() )
for x in iter:
yield VCFRecord( x, self )
def get_contigs(fns):
contigs = set()
i = 0
vcfs = []
for v in fns:
tbx = pysam.TabixFile(v, parser=pysam.asVCF())
cs = tbx.contigs
contigs.update(cs)
vcfs.append(tbx)
return (sorted(contigs, key=cmp_to_key(chrom_cmp)), vcfs)
def records(self, chromosome_num, start_pos_bp, end_pos_bp, parser=pysam.asVCF()):
'''
Returns an iterator for the file records.
'''
start_pos_bp = 1 if start_pos_bp is None else start_pos_bp
end_pos_bp = self.clens[str(chromosome_num)] if end_pos_bp is None else end_pos_bp
# subtract one since pysam uses incorrect 0-indexed positions
records = self.tabix_file.fetch( str(chromosome_num), start_pos_bp+self.indexDelta, end_pos_bp+self.indexDelta, parser)
return records
def records(self, chromosome_num, start_pos_bp, end_pos_bp, parser=pysam.asVCF()):
'''
Returns an iterator for the file records.
'''
start_pos_bp = 1 if start_pos_bp is None else start_pos_bp
end_pos_bp = self.clens[str(chromosome_num)] if end_pos_bp is None else end_pos_bp
# subtract one since pysam uses incorrect 0-indexed positions
records = self.tabix_file.fetch( str(chromosome_num), start=start_pos_bp+self.indexDelta, end=end_pos_bp+self.indexDelta, parser=parser)
return records
def check_nth_sample(options, genotype):
parser = pysam.asVCF()
for contig in genotype.contigs:
for record in text.parse_lines(genotype.fetch(contig), parser):
if len(record) <= options.nth_sample:
sys.stderr.write("ERROR: Sample %i selected with --nth-sample,"
" but file only contains %i sample(s)!\n"
% (options.nth_sample + 1, len(record)))
return False
return True
return True
def check_nth_sample(options, genotype):
parser = pysam.asVCF()
for contig in genotype.contigs:
for record in genotype.fetch(contig, parser=parser):
if len(record) <= options.nth_sample:
sys.stderr.write("ERROR: Sample %i selected with --nth-sample,"
" but file only contains %i sample(s)!\n" %
(options.nth_sample + 1, len(record)))
return False
return True
return True
def _read_files(filenames, args):
in_header = True
has_filters = False
vcf_parser = pysam.asVCF()
for line in fileinput.input(filenames):
if not line.startswith("#"):
in_header = False
line = line.rstrip("\n\r")
yield vcf_parser(line, len(line))
elif in_header:
if not (line.startswith("##") or has_filters):
has_filters = True
for item in sorted(vcffilter.describe_filters(args).items()):
print('##FILTER=<ID=%s,Description="%s">' % item)
print(line, end="")
def __init__(self, inFile, ploidy=1, parser=pysam.asVCF()):
TabixReader.__init__(self, inFile, parser=parser)
assert ploidy in (1,2)
self.ploidy = ploidy
self.clens = []
self.sample_names = None
for line in self.header:
if line.startswith('##contig=<ID=') and line.endswith('>'):
line = line[13:-1]
c = line.split(',')[0]
clen = int(line.split('=')[1])
self.clens.append((c,clen))
elif line.startswith('#CHROM'):
row = line.split('\t')
self.sample_names = row[9:]
self.clens = dict(self.clens)
assert self.sample_names
def __init__(self, inFile, ploidy=1, parser=pysam.asVCF()):
TabixReader.__init__(self, inFile, parser=parser)
assert ploidy in (1, 2)
self.ploidy = ploidy
self.clens = []
self.sample_names = None
for line in self.header:
if line.startswith('##contig=<ID=') and line.endswith('>'):
line = line[13:-1]
c = line.split(',')[0]
clen = int(line.split('=')[1])
self.clens.append((c, clen))
elif line.startswith('#CHROM'):
row = line.split('\t')
self.sample_names = row[9:]
self.clens = dict(self.clens)
assert self.sample_names
def testRead(self):
ncolumns = len(self.columns)
for x, r in enumerate(self.tabix.fetch(parser=pysam.asVCF())):
c = self.compare[x]
for y, field in enumerate(self.columns):
if field == "pos":
self.assertEqual(int(c[y]) - 1, getattr(r, field))
self.assertEqual(int(c[y]) - 1, r.pos)
else:
self.assertEqual( c[y], getattr( r, field ),
"mismatch in field %s: %s != %s" %\
( field,c[y], getattr( r, field ) ) )
self.assertEqual(len(c), len(r) + ncolumns)
for y in range(len(c) - ncolumns):
self.assertEqual(c[ncolumns + y], r[y])
def testRead( self ):
ncolumns = len(self.columns)
for x, r in enumerate(self.tabix.fetch( parser = pysam.asVCF() )):
c = self.compare[x]
for y, field in enumerate( self.columns ):
if field == "pos":
self.assertEqual( int(c[y]) - 1, getattr( r, field ) )
self.assertEqual( int(c[y]) - 1, r.pos )
else:
self.assertEqual( c[y], getattr( r, field ),
"mismatch in field %s: %s != %s" %\
( field,c[y], getattr( r, field ) ) )
self.assertEqual( len(c), len( r ) + ncolumns )
for y in range(len(c) - ncolumns):
self.assertEqual( c[ncolumns+y], r[y] )
def testWrite(self):
ncolumns = len(self.columns)
for x, r in enumerate(self.tabix.fetch(parser=pysam.asVCF())):
c = self.compare[x]
# check unmodified string
cmp_string = str(r)
ref_string = "\t".join([x for x in c])
self.assertEqual(ref_string, cmp_string)
# set fields and compare field-wise
for y, field in enumerate(self.columns):
# it is ok to have a missing format column
if y == 8 and y == len(c):
continue
if field == "pos":
rpos = getattr(r, field)
self.assertEqual(int(c[y]) - 1, rpos)
self.assertEqual(int(c[y]) - 1, r.pos)
# increment pos by 1
setattr(r, field, rpos + 1)
self.assertEqual(getattr(r, field), rpos + 1)
c[y] = str(int(c[y]) + 1)
else:
setattr(r, field, "test_%i" % y)
c[y] = "test_%i" % y
self.assertEqual(
c[y], getattr(r, field),
"mismatch in field %s: %s != %s" %
(field, c[y], getattr(r, field)))
if len(c) == 8:
self.assertEqual(0, len(r))
else:
self.assertEqual(len(c), len(r) + ncolumns)
for y in range(len(c) - ncolumns):
c[ncolumns + y] = "test_%i" % y
r[y] = "test_%i" % y
self.assertEqual(c[ncolumns + y], r[y])
def _get_hits(coords, annotation, parser_type):
"""Retrieve BED information, recovering if BED annotation file does have a chromosome.
"""
if parser_type == "bed":
parser = pysam.asBed()
elif parser_type == "vcf":
parser = pysam.asVCF()
elif parser_type == "tuple":
parser = pysam.asTuple()
elif parser_type is None:
parser = None
else:
raise ValueError("Unexpected parser type: %s" % parser)
chrom, start, end = coords
try:
hit_iter = annotation.fetch(str(chrom), start, end, parser=parser)
# catch invalid region errors raised by ctabix
except ValueError:
hit_iter = []
return hit_iter
def _get_hits(coords, annotation, parser_type):
"""Retrieve BED information, recovering if BED annotation file does have a chromosome.
"""
if parser_type == "bed":
parser = pysam.asBed()
elif parser_type == "vcf":
parser = pysam.asVCF()
elif parser_type == "tuple":
parser = pysam.asTuple()
elif parser_type is None:
parser = None
else:
raise ValueError("Unexpected parser type: %s" % parser)
chrom, start, end = coords
try:
hit_iter = annotation.fetch(str(chrom), start, end, parser=parser)
# catch invalid region errors raised by ctabix
except ValueError:
hit_iter = []
return hit_iter
def GetSumOfDifferencesFromTheReference(vcfpath):
from subprocess import check_call
from utilBMF.HTSUtils import TrimExt
import pysam
import numpy as np
from sys import stderr
from itertools import chain
cfi = chain.from_iterable
bgvcfpath = TrimExt(vcfpath) + ".gz"
check_call("bgzip -c %s > %s" % (vcfpath, bgvcfpath), shell=True)
stderr.write("bgvcf now at %s" % bgvcfpath)
tabixstr = "tabix " + bgvcfpath
stderr.write("Now calling tabixstr: '%s'" % tabixstr)
check_call("tabix %s" % bgvcfpath, shell=True)
infh = open(bgvcfpath, "rb")
tabixhandle = pysam.tabix_iterator(infh, pysam.asVCF())
return np.sum(np.array(list(cfi([dict(tup.split("=") for
tup in i.info.split(";"))[
'I16'].split(",")[2:4] for i in tabixhandle if
"INDEL" not in i.info])), dtype=np.int64))
def testWrite(self):
ncolumns = len(self.columns)
for x, r in enumerate(self.tabix.fetch(parser=pysam.asVCF())):
c = self.compare[x]
# check unmodified string
cmp_string = str(r)
ref_string = "\t".join([x for x in c])
self.assertEqual(ref_string, cmp_string)
# set fields and compare field-wise
for y, field in enumerate(self.columns):
# it is ok to have a missing format column
if y == 8 and y == len(c):
continue
if field == "pos":
rpos = getattr(r, field)
self.assertEqual(int(c[y]) - 1, rpos)
self.assertEqual(int(c[y]) - 1, r.pos)
# increment pos by 1
setattr(r, field, rpos + 1)
self.assertEqual(getattr(r, field), rpos + 1)
c[y] = str(int(c[y]) + 1)
else:
setattr(r, field, "test_%i" % y)
c[y] = "test_%i" % y
self.assertEqual(c[y], getattr(r, field),
"mismatch in field %s: %s != %s" %
(field, c[y], getattr(r, field)))
if len(c) == 8:
self.assertEqual(0, len(r))
else:
self.assertEqual(len(c), len(r) + ncolumns)
for y in range(len(c) - ncolumns):
c[ncolumns + y] = "test_%i" % y
r[y] = "test_%i" % y
self.assertEqual(c[ncolumns + y], r[y])
def select_vcf_records(bed_records, vcf_records):
"""Returns an iterable of VCF records, corresponding to the contents of each
region specified by the BED records. Records are returned at most once, even
if covered by multiple BED records."""
contigs = frozenset(vcf_records.contigs)
vcf_parser = pysam.asVCF()
# Timer class used processing progress; meant primarily for BAM files
progress = timer.BAMTimer(None)
# Cache of positions observed for this contig, to prevent returning
# positions in overlapping regions multiple times
contig_cache = None
contig_cache_name = None
for bed in sorted(bed_records):
if bed.contig not in contigs:
# Skip contigs for which no calls have been made (e.g. due to
# low coverage. Otherwise Pysam raises an exception.
continue
elif contig_cache_name != bed.contig:
# Reset cache per contig, to save memory
contig_cache = set()
contig_cache_name = bed.contig
for record in vcf_records.fetch(bed.contig,
bed.start,
bed.end,
parser=vcf_parser):
progress.increment()
if record.pos in contig_cache:
# We've already reported this VCF record
continue
contig_cache.add(record.pos)
# Skip records filtered by VCF_filter
if record.filter in ('.', "PASS"):
yield record
progress.finalize()
def _read_files(args):
in_header = True
has_filters = False
vcf_parser = pysam.asVCF()
for filename in args.filenames:
with open_ro(filename, "rb") as handle:
for line in handle:
if not line.startswith(b"#"):
in_header = False
line = line.rstrip(b"\n\r")
vcf = vcf_parser(line, len(line))
if args.reset_filter:
vcf.filter = "."
yield vcf
elif in_header:
if not (line.startswith(b"##") or has_filters):
has_filters = True
for item in sorted(
vcffilter.describe_filters(args).items()):
print('##FILTER=<ID=%s,Description="%s">' % item)
print(line.decode("utf-8"), end="")
def __init__(self, inFile, ploidy=1, parser=pysam.asVCF()):
# using old-style class superclass calling here
# since TabixReader is derived from pysam.TabixFile
# which is itself an old-style class (due to Cython version?)
TabixReader.__init__(self, inFile, parser=parser)
# when pysam uses new-style classes, we can replace with:
#super(VcfReader, self).__init__(inFile, parser=parser)
assert ploidy in (1, 2)
self.ploidy = ploidy
self.clens = []
self.sample_names = None
for line in self.header:
line = bytes_to_string(line)
if line.startswith('##contig=<ID=') and line.endswith('>'):
line = line[13:-1]
c = line.split(',')[0]
clen = int(line.split('=')[1])
self.clens.append((c, clen))
elif line.startswith('#CHROM'):
row = line.split('\t')
self.sample_names = row[9:]
self.clens = dict(self.clens)
assert self.sample_names
def __init__(self, inFile, ploidy=1, parser=pysam.asVCF()):
# using old-style class superclass calling here
# since TabixReader is derived from pysam.TabixFile
# which is itself an old-style class (due to Cython version?)
TabixReader.__init__(self, inFile, parser=parser)
# when pysam uses new-style classes, we can replace with:
#super(VcfReader, self).__init__(inFile, parser=parser)
assert ploidy in (1, 2)
self.ploidy = ploidy
self.clens = []
self.sample_names = None
for line in self.header: # pylint: disable=E1101
line = bytes_to_string(line)
if line.startswith('##contig=<ID=') and line.endswith('>'):
line = line[13:-1]
c = line.split(',')[0]
clen = int(line.split('=')[1])
self.clens.append((c, clen))
elif line.startswith('#CHROM'):
row = line.split('\t')
self.sample_names = row[9:]
self.clens = dict(self.clens)
assert self.sample_names
def select_vcf_records(bed_records, vcf_records):
"""Returns an iterable of VCF records, corresponding to the contents of each
region specified by the BED records. Records are returned at most once, even
if covered by multiple BED records."""
contigs = frozenset(vcf_records.contigs)
vcf_parser = pysam.asVCF()
# Timer class used processing progress; meant primarily for BAM files
progress = timer.BAMTimer(None)
# Cache of positions observed for this contig, to prevent returning
# positions in overlapping regions multiple times
contig_cache = None
contig_cache_name = None
for bed in sorted(bed_records):
if bed.contig not in contigs:
# Skip contigs for which no calls have been made (e.g. due to
# low coverage. Otherwise Pysam raises an exception.
continue
elif contig_cache_name != bed.contig:
# Reset cache per contig, to save memory
contig_cache = set()
contig_cache_name = bed.contig
for record in vcf_records.fetch(bed.contig, bed.start, bed.end, parser = vcf_parser):
progress.increment()
if record.pos in contig_cache:
# We've already reported this VCF record
continue
contig_cache.add(record.pos)
# Skip records filtered by VCF_filter
if record.filter in ('.', "PASS"):
yield record
progress.finalize()
def RunTools(args):
cfg = Parse.generate_tools_cfg(args)
Parse.print_tools_options(cfg)
if not cfg['debug']:
logging.disable(logging.CRITICAL)
regions_df = pd.read_table(cfg['region_file'], compression='gzip' if cfg['region_file'].split('.')[-1] == 'gz' else None)
regions_df = regions_df[regions_df['job'] == int(cfg['job'])].reset_index(drop=True)
return_values = {}
print ''
print "initializing out file"
try:
bgzfile = bgzf.BgzfWriter(cfg['out'] + '.gz', 'wb')
except:
print Process.Error("failed to initialize bgzip format out file " + cfg['out'] + '.gz').out
return 1
if cfg['cpus'] > 1:
pool = mp.Pool(cfg['cpus']-1)
for i in xrange(1,cfg['cpus']):
return_values[i] = pool.apply_async(process_regions, args=(regions_df,cfg,i,True,))
print "submitting job on cpu " + str(i) + " of " + str(cfg['cpus'])
pool.close()
print "executing job for cpu " + str(cfg['cpus']) + " of " + str(cfg['cpus']) + " via main process"
main_return = process_regions(regions_df,cfg,cfg['cpus'],True)
pool.join()
if 1 in [return_values[i].get() for i in return_values] or main_return == 1:
print Process.Error("error detected, see log files").out
return 1
else:
main_return = process_regions(regions_df,cfg,1,True)
if main_return == 1:
print Process.Error("error detected, see log files").out
return 1
for i in xrange(1,cfg['cpus']+1):
try:
logfile = open(cfg['out'] + '.cpu' + str(i) + '.log', 'r')
except:
print Process.Error("failed to initialize log file " + cfg['out'] + '.cpu' + str(i) + '.log').out
return 1
print logfile.read()
logfile.close()
os.remove(cfg['out'] + '.cpu' + str(i) + '.log')
written = False
for i in xrange(1,cfg['cpus']+1):
cpu_regions_df = regions_df[regions_df['cpu'] == i].reset_index()
for j in xrange(0,len(cpu_regions_df.index)):
f_temp=glob.glob(cfg['out'] + '.cpu' + str(i) + '.chr' + cpu_regions_df['region'][j].replace(':','bp') + '*.gz')[0]
try:
h=pysam.TabixFile(filename=f_temp,parser=pysam.asVCF())
except:
print Process.Error("failed to load vcf file " + f_temp)
return 1
if not written:
for row in h.header:
bgzfile.write(str(row) + '\n')
written = True
h_iter = h.fetch(region=str(cpu_regions_df['chr'][j]))
for row in h_iter:
bgzfile.write(str(row) + '\n')
for f in glob.glob(cfg['out'] + '.cpu' + str(i) + '.chr' + cpu_regions_df['region'][j].replace(':','bp') + '.*'):
os.remove(f)
bgzfile.close()
print "indexing out file"
try:
pysam.tabix_index(cfg['out'] + '.gz',preset="vcf",force=True)
except:
print Process.Error('failed to generate index').out
return 1
print "process complete"
return 0
def process_piece(merge_info):
stats = {}
try:
output_file_left = None
output_file_right = None
if merge_info.output_file_left:
output_file_left = open(merge_info.output_file_left, "w")
if merge_info.output_file_right:
output_file_right = open(merge_info.output_file_right, "w")
mi = ['L']
if merge_info.diploid:
mi = ['L', 'R']
LOG.info("Processing Chromosome {0}...".format(merge_info.chromosome))
iterators = []
discard_functions = []
for i, file_info in enumerate(merge_info.vcf_files):
vcf_tabix = pysam.TabixFile(file_info.file_name)
try:
vcf_iterator = vcf_tabix.fetch(merge_info.chromosome,
parser=pysam.asVCF())
iterators.append(vcf_iterator)
except ValueError as ve:
iterators.append(None)
if file_info.discard_file:
vcf_discard = open(file_info.discard_file, "w")
def discard_record(rec):
vcf_discard.write(str(rec))
vcf_discard.write("\n")
discard_functions.append(discard_record)
else:
discard_functions.append(lambda rec: None)
n = 0
line_numbers = 0
for vcf_records in walk_vcfs_together(iterators):
for i, vcf_record in enumerate(vcf_records):
#LOG.debug(vcf_record)
if vcf_record is None:
continue
#LOG.debug(vcf_record.alt)
#LOG.debug(type(vcf_record.alt))
gt = vcf.parse_gt_tuple(vcf_record,
merge_info.vcf_files[i].sample_index)
#LOG.debug(gt)
line_numbers = line_numbers + 1
if gt.is_snp:
# snp
if merge_info.passed and 'PASS' not in vcf_record.filter:
discard_functions[i](vcf_record)
#LOG.debug("Processing SNP {}".format(vcf_record))
n += 1
if merge_info.quality and gt.fi == '0':
discard_functions[i](vcf_record)
elif gt.left is None or gt.right is None:
discard_functions[i](vcf_record)
else:
if merge_info.diploid:
# 0 i sthe same as REF and do not need
if gt.gt_left != 0:
output_file_left.write(
"{}_L\t{}\t{}\t{}\t{}\t{}\n".format(
merge_info.chromosome,
vcf_record.pos + 1, '.',
vcf_record.ref, gt.left, '.'))
if gt.gt_right != 0:
output_file_right.write(
"{}_R\t{}\t{}\t{}\t{}\t{}\n".format(
merge_info.chromosome,
vcf_record.pos + 1, '.',
vcf_record.ref, gt.right, '.'))
else:
if gt.gt_left == gt.gt_right and gt.gt_left != 0:
# ignore heterozygotes 0/1, 1/0, only process 0/0 and 1/1
#LOG.debug("ACCEPTED")
#LOG.debug('pos {} : ref {}, left {}, right {}'.format(vcf_snp.pos, vcf_snp.ref, gt.left, gt.right))
output_file_left.write(
"{}\t{}\t{}\t{}\t{}\t{}\n".format(
merge_info.chromosome,
vcf_record.pos + 1, '.',
vcf_record.ref, gt.left, '.'))
else:
# indel
LOG.debug("Processing INDEL {}".format(vcf_record))
if merge_info.passed and 'PASS' not in vcf_record.filter:
LOG.debug("TOSSED: FILTERED ON PASS")
LOG.debug(vcf_record)
stats = update_stats(stats, 'FILTERED ON PASS')
discard_functions[i](vcf_record)
continue
elif merge_info.quality and gt.fi == '0':
# FI : Whether a sample was a Pass(1) or fail (0) based on FILTER values
LOG.debug("TOSSED: FILTERED ON QUALITY")
LOG.debug(vcf_record)
stats = update_stats(stats, 'FILTERED ON QUALITY')
discard_functions[i](vcf_record)
continue
elif gt.left is None and gt.right is None:
LOG.debug("TOSSED: NO STRAIN DATA")
LOG.debug(vcf_record)
stats = update_stats(stats, 'NO STRAIN DATA')
LOG.debug(i)
LOG.debug(type(vcf_record))
discard_functions[i](vcf_record)
continue
elif not merge_info.diploid and gt.left != gt.right:
# haploid or hexaploid
# gt must be equal
LOG.debug("TOSSED: HETEROZYGOUS")
LOG.debug(vcf_record)
stats = update_stats(stats, 'HETEROZYGOUS')
discard_functions[i](vcf_record)
continue
# START L AND R, ONLY R IF DIPLOID
for l_or_r in mi:
#LOG.debug("******************")
#LOG.debug(l_or_r)
lr_out = ''
if l_or_r == 'L':
#LOG.debug("->LEFT")
lr_out = '_L' if merge_info.diploid else ''
alt_seq = str(gt.left)
stats = merge_info.stats_left
output_file = output_file_left
prev_next_ref_pos = merge_info.prev_next_ref_pos_left
else:
#LOG.debug("->RIGHT")
lr_out = '_R' if merge_info.diploid else ''
alt_seq = str(gt.right)
stats = merge_info.stats_right
output_file = output_file_right
prev_next_ref_pos = merge_info.prev_next_ref_pos_right
LOG.debug(
"prev_next_ref_pos={}".format(prev_next_ref_pos))
if gt.ref == alt_seq:
LOG.debug("TOSSED, REF AND ALT ARE EQUAL")
LOG.debug(vcf_record)
stats = update_stats(stats,
'REF AND ALT ARE EQUAL')
discard_functions[i](vcf_record)
continue
orig_alt_seq = alt_seq
LOG.debug("SAMPLE: {0}".format(
vcf_record[merge_info.vcf_files[i].sample_index]))
LOG.debug(
"REF='{0}', ALT_L='{1}', ALT_R='{2}'. POS={3}".
format(gt.ref, gt.left, gt.right, vcf_record.pos))
position = vcf_record.pos + 1
ref_seq = str(gt.ref)
len_ref = len(ref_seq)
len_alt = len(alt_seq)
base_changes = len_ref - len_alt
base_pos_diff = 0
if position < prev_next_ref_pos:
LOG.debug(
"TOSSED: VCF ROLLBACK: {0}".format(vcf_record))
LOG.debug(vcf_record)
stats = update_stats(stats, 'VCF ROLLBACK')
discard_functions[i](vcf_record)
continue
# find the position where the first base change is
for n in xrange(min(len_ref, len_alt)):
if ref_seq[n] != alt_seq[n]:
base_pos_diff = n
break
# if it is 0, take the minimum length
if base_pos_diff == 0:
base_pos_diff = min(len_ref, len_alt)
# add the base position difference
position += base_pos_diff
# recalculate the strings
shared_bases = ref_seq[:base_pos_diff]
ref_seq = ref_seq[base_pos_diff:]
alt_seq = alt_seq[base_pos_diff:]
dt = len(ref_seq)
dq = len(alt_seq)
next_ref_pos = position + len(ref_seq)
fragment_size = position - prev_next_ref_pos
'''
LOG.debug(' gt.ref: {0}'.format(gt.ref))
LOG.debug(' ref_seq: {0}'.format(ref_seq))
LOG.debug(' dt: {0}'.format(dt))
LOG.debug(' gt.alt: {0}'.format(orig_alt_seq))
LOG.debug(' alt_seq: {0}'.format(alt_seq))
LOG.debug(' dq: {0}'.format(dq))
LOG.debug(' position: {0}'.format(position))
LOG.debug('prev_next_ref_pos: {0}'.format(prev_next_ref_pos))
LOG.debug(' next_ref_pos: {0}'.format(next_ref_pos))
LOG.debug(' fragment_size: {0}'.format(fragment_size))
LOG.debug(' base_changes: {0}'.format(base_changes))
LOG.debug(' base_pos_diff: {0}'.format(base_pos_diff))
LOG.debug(' shared_bases: {0}'.format(shared_bases))
'''
# fix any 0 length
if fragment_size < 0:
#LOG.debug("TOSSED: FRAGMENT: {0}".format(vcf_record))
stats = update_stats(stats, 'FRAGMENT SIZE < 0')
discard_functions[i](vcf_record)
continue
if fragment_size != 0:
ref_str = ref_seq if ref_seq else '.'
alt_str = alt_seq if alt_seq else '.'
out = "{}{}\t{}\t{}\t{}\t{}\t{}\n".format(
merge_info.chromosome, lr_out,
vcf_record.pos + 1, shared_bases, ref_str,
alt_str, fragment_size)
LOG.debug(out)
output_file.write(out)
else:
#
# THIS SHOULD NOT HAPPEN
#
raise exceptions.G2GVCFError(
'Conflicting VCF entries')
stats = update_stats(stats, 'ACCEPTED')
if l_or_r == 'L':
merge_info.stats_left = stats
merge_info.prev_next_ref_pos_left = next_ref_pos
LOG.debug(
'setting merge_info.prev_next_ref_pos_left={}'.
format(merge_info.prev_next_ref_pos_left))
else:
merge_info.stats_right = stats
merge_info.prev_next_ref_pos_right = next_ref_pos
LOG.debug(
'setting merge_info.prev_next_ref_pos_right={}'
.format(merge_info.prev_next_ref_pos_right))
if merge_info.output_file_left:
output_file_left.close()
if merge_info.output_file_right:
output_file_right.close()
except KeyboardInterrupt:
raise exceptions.KeyboardInterruptError()
except Exception as e:
g2g_utils._show_error()
raise Exception("Unknown exception")
ret = {}
ret['chrom'] = merge_info.chromosome
ret['stats'] = stats
ret['merge_info'] = merge_info
ret['line_numbers'] = line_numbers
return ret
except IOError:
sys.exit("Gemini cannot open this annotation file: %s. \n"
"Have you installed the annotation files? If so, "
"have they been moved or deleted? Exiting...\n\n"
"For more details:\n\t"
"http://gemini.readthedocs.org/en/latest/content/"
"#installation.html\#installing-annotation-files\n" %
anno_files[anno])
# ## Standard access to Tabix indexed files
PARSERS = {
"bed": pysam.asBed(),
"vcf": pysam.asVCF(),
"tuple": pysam.asTuple(),
None: None
}
def _get_hits(coords, annotation, parser_type, _parsers=PARSERS):
"""Retrieve BED information, recovering if BED annotation file does have a chromosome.
"""
try:
parser = _parsers[parser_type]
except KeyError:
raise ValueError("Unexpected parser type: %s" % parser)
chrom, start, end = coords
try:
hit_iter = annotation.fetch(str(chrom), start, end, parser=parser)
def process_piece(filename_vcf, chrom, chrom_length, sample_index, chain_info, diploid, passed, quality, vcf_keep, vcf_discard_file):
ret = {'chrom': chrom, 'stats': {}, 'chain_info': chain_info}
stats = OrderedDict()
stats['ACCEPTED'] = 0
if vcf_keep:
vcf_discard = open(vcf_discard_file, "w")
line_no = 0
try:
LOG.info("Processing Chromosome {0}...".format(chrom))
tb = pysam.TabixFile(filename_vcf)
for vcf_rec in tb.fetch(chrom, parser=pysam.asVCF()):
line_no += 1
try:
gt = parse_gt_new(vcf_rec, sample_index)
except:
LOG.info("Unable to parse record, improper VCF file?")
continue
LOG.debug('\n')
LOG.debug(vcf_rec)
LOG.debug(gt)
LOG.debug(vcf_rec[sample_index])
if passed and 'PASS' not in vcf_rec.FILTER:
LOG.debug("TOSSED: FILTERED ON PASS")
stats = update_stats(stats, 'FILTERED ON PASS')
if vcf_keep:
vcf_discard.write(vcf_rec)
vcf_discard.write("\n")
continue
elif quality and gt.fi == '0':
# FI : Whether a sample was a Pass(1) or fail (0) based on FILTER values
LOG.debug("TOSSED: FILTERED ON QUALITY")
stats = update_stats(stats, 'FILTERED ON QUALITY')
if vcf_keep:
vcf_discard.write(vcf_rec)
vcf_discard.write("\n")
continue
elif gt.left is None and gt.right is None:
LOG.debug("TOSSED: NOT RELEVANT")
stats = update_stats(stats, 'NOT RELEVANT')
if vcf_keep:
vcf_discard.write(vcf_rec)
vcf_discard.write("\n")
continue
elif not diploid and gt.left != gt.right:
# haploid or hexaploid
# gt must be equal
LOG.debug("TOSSED: HETEROZYGOUS")
stats = update_stats(stats, 'HETEROZYGOUS')
if vcf_keep:
vcf_discard.write(vcf_rec)
vcf_discard.write("\n")
continue
# START L AND R, ONLY R IF DIPLOID
for ci, lr in chain_info.iteritems():
if ci == 'left':
alt_seq = str(gt.left)
else:
alt_seq = str(gt.right)
if gt.ref == alt_seq:
LOG.debug("TOSSED, SAME AS REF")
lr.stats = update_stats(lr.stats, 'SAME AS REF')
if vcf_keep:
vcf_discard.write(vcf_rec)
vcf_discard.write("\n")
continue
orig_alt_seq = alt_seq
LOG.debug("SAMPLE: {0}".format(vcf_rec[sample_index]))
LOG.debug("REF='{0}', ALT_L='{1}', ALT_R='{2}'. POS={3}".format(gt.ref, gt.left, gt.right, vcf_rec.pos))
position = vcf_rec.pos + 1
ref_seq = str(gt.ref)
len_ref = len(ref_seq)
len_alt = len(alt_seq)
base_changes = len_ref - len_alt
base_pos_diff = 0
if position < lr.prev_next_ref_pos:
LOG.debug("TOSSED: CONFLICTING VCF ENTRIES: {0}".format(vcf_rec))
lr.stats = update_stats(lr.stats, 'CONFLICTING VCF ENTRIES')
if vcf_keep:
vcf_discard.write(vcf_rec)
vcf_discard.write("\n")
continue
# find the position where the first base change is
for n in xrange(min(len_ref, len_alt)):
if ref_seq[n] != alt_seq[n]:
base_pos_diff = n
break
# if it is 0, take the minimum length
if base_pos_diff == 0:
base_pos_diff = min(len_ref, len_alt)
# add the base position difference
position += base_pos_diff
# recalculate the strings
shared_bases = ref_seq[:base_pos_diff]
ref_seq = ref_seq[base_pos_diff:]
alt_seq = alt_seq[base_pos_diff:]
dt = len(ref_seq)
dq = len(alt_seq)
next_ref_pos = position + len(ref_seq)
fragment_size = position - lr.prev_next_ref_pos
LOG.debug(' gt.ref: {0}'.format(gt.ref))
LOG.debug(' ref_seq: {0}'.format(ref_seq))
LOG.debug(' dt: {0}'.format(dt))
LOG.debug(' gt.alt: {0}'.format(orig_alt_seq))
LOG.debug(' alt_seq: {0}'.format(alt_seq))
LOG.debug(' dq: {0}'.format(dq))
LOG.debug(' position: {0}'.format(position))
LOG.debug('prev_next_ref_pos: {0}'.format(lr.prev_next_ref_pos))
LOG.debug(' next_ref_pos: {0}'.format(next_ref_pos))
LOG.debug(' fragment_size: {0}'.format(fragment_size))
LOG.debug(' base_changes: {0}'.format(base_changes))
LOG.debug(' base_pos_diff: {0}'.format(base_pos_diff))
LOG.debug(' shared_bases: {0}'.format(shared_bases))
# fix any 0 length
if fragment_size < 0:
LOG.debug("TOSSED: CONFLICTING VCF ENTRIES: {0}".format(vcf_rec))
lr.stats = update_stats(lr.stats, 'CONFLICTING VCF ENTRIES')
if vcf_keep:
vcf_discard.write(vcf_rec)
vcf_discard.write("\n")
continue
if fragment_size != 0:
ref_str = ref_seq if ref_seq else '.'
alt_str = alt_seq if alt_seq else '.'
lr.chain_entries.append([fragment_size, len(ref_seq), len(alt_seq), shared_bases, ref_str, alt_str, vcf_rec.pos+1])
else:
#
# THIS SHOULD NOT HAPPEN
#
raise G2GChainFileError('Unable to create chain file due to conflicting VCF entries')
lr.stats = update_stats(lr.stats, 'ACCEPTED')
LOG.debug(lr.chain_entries[-1])
last_position = position
lr.prev_next_ref_pos = next_ref_pos
lr.sums[0] += fragment_size
lr.sums[1] += dt
lr.sums[2] += dq
prev_line = vcf_rec
#lr.prev_chrom = vcf_rec.contig
chain_info[ci] = lr
for ci, lr in chain_info.iteritems():
#LOG.debug("CHROMOSOME[{0}] LENGTH = {1}".format(lr.prev_chrom, chrom_length))
lr.chromosome = chrom
lr.chromosome_length = chrom_length
lr.last_fragment_size = chrom_length - lr.sums[0] - lr.sums[1]
lr.end_length = lr.sums[0] + lr.last_fragment_size + lr.sums[2]
lr.number_vcf_lines = line_no
chain_info[ci] = lr
if vcf_keep:
vcf_discard.close()
except KeyboardInterrupt:
raise KeyboardInterruptError()
except Exception, e:
pass
def process_vcf_into_selscan_tped(cls, vcf_file, gen_map_file, outfile_location,
outfile_prefix, chromosome_num, samples_to_include=None, start_pos_bp=None, end_pos_bp=None, ploidy=2,
consider_multi_allelic=True, rescale_genetic_distance=False, include_variants_with_low_qual_ancestral=False, coding_function=None,
multi_alleli_merge_function="AND"):
"""
Process a bgzipped-VCF (such as those included in the Phase 3 1000 Genomes release) into a gzip-compressed
tped file of the sort expected by selscan.
"""
assert ploidy > 0
processor = VCFReader(vcf_file)
end_pos = processor.clens[str(chromosome_num)] if end_pos_bp is None else end_pos_bp
records = processor.records( str(chromosome_num), start_pos_bp, end_pos, pysam.asVCF())
outTpedFile = outfile_location + "/" + outfile_prefix + ".tped.gz"
outTpedMetaFile = outfile_location + "/" + outfile_prefix + ".tped.allele_metadata.gz"
if samples_to_include is not None and len(samples_to_include) > 0:
indices_of_matching_samples = sorted([processor.sample_names.index(x) for x in samples_to_include])
else:
indices_of_matching_samples = list(range(0,len(processor.sample_names)))
indices_of_matching_genotypes = [(x*2, (x*2)+1) for x in indices_of_matching_samples]
indices_of_matching_genotypes = list(np.ravel(np.array(indices_of_matching_genotypes)))
rm = RecomMap(gen_map_file)
for filePath in [outTpedFile, outTpedMetaFile]:
assert not os.path.exists(filePath), "File {} already exists. Consider removing this file or specifying a different output prefix. Processing aborted.".format(filePath)
mergeOperatorString = ""
if multi_alleli_merge_function == "OR":
mergeOperatorString = "|"
if multi_alleli_merge_function == "AND":
mergeOperatorString = "&"
if multi_alleli_merge_function == "XOR":
mergeOperatorString = "^"
startTime = datetime.now()
sec_remaining_avg = 0
current_pos_bp = 1
with util.file.open_or_gzopen(outTpedFile, 'wt') as of1, util.file.open_or_gzopen(outTpedMetaFile, 'wt') as of2:
# WRITE header for metadata file here with selected subset of sample_names
headerString = "CHROM VARIANT_ID POS_BP MAP_POS_CM REF_ALLELE ALT_ALLELE ANCESTRAL_CALL ALLELE_FREQ_IN_POP\n".replace(" ","\t")
of2.write(headerString)
of1linesToWrite = []
of2linesToWrite = []
recordLengths = set() #
recordCount = 0
mostRecentRecordPosSeen = -1
positionHasBeenSeenBefore = False
previouslyCodedGenotypes = GenoRecord([])
lineToWrite1 = None
lineToWrite2 = None
previousAncestral = None
ancestralDiffersFromPrevious = True
for record in records:
# in some cases, there may be records with duplicate positions in the VCF file
# to account for that we collapse rows that pass our filter and then write out the rows when we
# encounter a record with a new position
if record.pos != mostRecentRecordPosSeen:
if positionHasBeenSeenBefore and not consider_multi_allelic:
lineToWrite1 = None
lineToWrite2 = None
if lineToWrite1 is not None and lineToWrite2 is not None:
if len(previouslyCodedGenotypes) == ploidy*len(indices_of_matching_samples):
# write the output line
of1.write(lineToWrite1)
of2.write(lineToWrite2)
else:
genotypesCount = len(previouslyCodedGenotypes)
countSpecificTpedName = outTpedFile.replace(outfile_prefix, outfile_prefix + "_" + str(genotypesCount))
countSpecificMetafileName = outTpedMetaFile.replace(outfile_prefix, outfile_prefix + "_" + str(genotypesCount))
with util.file.open_or_gzopen(countSpecificTpedName, 'at') as of1l, util.file.open_or_gzopen(countSpecificMetafileName, 'at') as of2l:
of1l.write(lineToWrite1)
of2l.write(lineToWrite2)
lineToWrite1 = None
lineToWrite2 = None
mostRecentRecordPosSeen = record.pos
positionHasBeenSeenBefore = False
else:
positionHasBeenSeenBefore = True
# if the variant is a SNP
# OLD style looking at INFO VT value:
# processor.variant_is_type(record.info, "SNP"):
VALID_BASES = ["A","C","G","T","a","c","g","t"] #"N","n"
if (len(record.ref) == 1 and len(record.alt) == 1) or ( all(variant in VALID_BASES for variant in record.ref.split(",")) and
all(variant in VALID_BASES for variant in record.alt.split(",")) ):
alternateAlleles = [record.alt]
if record.alt not in ['A','T','C','G']:
#print record.alt
if consider_multi_allelic:
pass
alternateAlleles = record.alt.split(",")
else:
# continue on to next variant record
continue
ancestral_allele = processor.parse_ancestral(record.info)
chromStr = "chr{}".format(record.contig)
# if the AA is populated, and the call meets the specified criteria
if (ancestral_allele in ['A','T','C','G']) or include_variants_with_low_qual_ancestral:#(include_variants_with_low_qual_ancestral and ancestral_allele in ['a','t','c','g',]):
if include_variants_with_low_qual_ancestral:
if ancestral_allele in ['a','t','c','g',]:
ancestral_allele = ancestral_allele.upper()
else: #if no info, encode ref as ancestral
ancestral_allele = record.ref
if previousAncestral != ancestral_allele:
previousAncestral = ancestral_allele
ancestralDiffersFromPrevious = True
else:
ancestralDiffersFromPrevious = False
recordString = record.__str__()
match = cls.genoRegex.match(recordString)
if match:
rawGenos = match.group("genos")
genos = rawGenos[::2]
recordPosition = record.pos+1
if chromosome_num.upper() != "X":
try:
genotypes_for_selected_samples = operator.itemgetter(*indices_of_matching_genotypes)(genos)
except IndexError:
# if something about this locus makes the genotype line incomplete
# we can continue to the next record. This can happen if '.' is incorrectly used in place of '.|.' in phased VCFs.
# anything that causes the genotype line length to mismatch with the number of genotypes
print("\ngenotype line length appears to be off:")
print(record)
continue
except Exception: # if this is a record of mixed ploidy, that is to say the X chromosome
# raise an exception
raise
else:
matching_genotypes = np.array(record[:len(record)])[indices_of_matching_samples]
genotypes_for_selected_samples_split = [x.split("|") for x in matching_genotypes]
genotypes_for_selected_samples = [y for x in genotypes_for_selected_samples_split for y in x]
recordLengths.add(len(genotypes_for_selected_samples))
map_pos_cm = rm.physToMap(chromStr, record.pos, rescale=rescale_genetic_distance)
numberOfHaplotypes = float(len(genotypes_for_selected_samples))
codingFunc = np.vectorize(coding_function)
coded_genotypes_for_selected_samples = GenoRecord(["0"] * len(genotypes_for_selected_samples))
if consider_multi_allelic:
#coded_genotypes_for_selected_samples = GenoRecord(["1"] * len(genotypes_for_selected_samples))
for idx, altAllele in enumerate(alternateAlleles):
value_of_current_allele = str(idx+1)
coded_genotypes_for_selected_samples_for_allele = GenoRecord(codingFunc(genotypes_for_selected_samples, record.ref, altAllele, ancestral_allele, value_of_current_allele))
#coded_genotypes_for_selected_samples |= coded_genotypes_for_selected_samples_for_allele
if idx==0:
coded_genotypes_for_selected_samples = coded_genotypes_for_selected_samples_for_allele
else:
coded_genotypes_for_selected_samples = coded_genotypes_for_selected_samples.f[mergeOperatorString](coded_genotypes_for_selected_samples_for_allele)
#coded_genotypes_for_selected_samples = np.array(list(str(bin(int("".join(coded_genotypes_for_selected_samples),2) | int("".join(coded_genotypes_for_selected_samples_for_allele),2)))[2:].zfill(len(coded_genotypes_for_selected_samples))))
else:
coded_genotypes_for_selected_samples = GenoRecord(codingFunc(genotypes_for_selected_samples, record.ref, record.alt, ancestral_allele, "1"))
# if this is the first record in the file, create an array filled with zeros for the previously coded alleles
if recordCount == 0:
previouslyCodedGenotypes = GenoRecord(["1"] * len(genotypes_for_selected_samples))
# bitwise OR coded genotypes for duplicate records, merge variants
# ...except selscan logic is inverted, so bitwise AND
# TODO: invert?
# record @ pos1 = 001001
# record @ pos1 = 100001
# -------
# coded result = 101001
#log.debug(genotypes_for_selected_samples)
#log.debug(coded_genotypes_for_selected_samples)
if positionHasBeenSeenBefore:
#coded_genotypes_for_selected_samples |= previouslyCodedGenotypes
coded_genotypes_for_selected_samples = coded_genotypes_for_selected_samples.f[mergeOperatorString](previouslyCodedGenotypes)
#coded_genotypes_for_selected_samples = np.array(list(str(bin(int("".join(coded_genotypes_for_selected_samples),2) & int("".join(previouslyCodedGenotypes),2)))[2:].zfill(len(coded_genotypes_for_selected_samples))))
previouslyCodedGenotypes = coded_genotypes_for_selected_samples
allele_freq_for_pop = float(list(coded_genotypes_for_selected_samples).count("1")) / numberOfHaplotypes
outStrDict = cls._build_variant_output_strings(record.contig, str(1), recordPosition,
map_pos_cm, coded_genotypes_for_selected_samples, record.ref, record.alt,
ancestral_allele, allele_freq_for_pop)
lineToWrite1 = outStrDict["tpedString"]
lineToWrite2 = outStrDict["metadataString"].replace(" ","\t")
recordCount += 1
current_pos_bp = int(recordPosition)
if recordCount % 1000 == 0:
number_of_seconds_elapsed = (datetime.now() - startTime).total_seconds()
bp_per_sec = float(current_pos_bp) / float(number_of_seconds_elapsed)
bp_remaining = end_pos - current_pos_bp
sec_remaining = bp_remaining / bp_per_sec
sec_remaining_avg = cls._moving_avg(sec_remaining, sec_remaining_avg, 10)
time_left = timedelta(seconds=sec_remaining_avg)
if sec_remaining > 10:
human_time_remaining = relative_time(datetime.utcnow()+time_left)
print("")
print(("Completed: {:.2%}".format(float(current_pos_bp)/float(end_pos))))
print(("Estimated time of completion: {}".format(human_time_remaining)))
#log.info("Genotype counts found: %s", str(list(recordLengths)))
if positionHasBeenSeenBefore and not consider_multi_allelic:
lineToWrite1 = None
lineToWrite2 = None
if lineToWrite1 is not None and lineToWrite2 is not None:
if len(previouslyCodedGenotypes) == ploidy*len(indices_of_matching_samples):
# write the output lines
of1.write(lineToWrite1)
of2.write(lineToWrite2)
else:
genotypesCount = len(previouslyCodedGenotypes)
countSpecificTpedName = outTpedFile.replace(outfile_prefix, outfile_prefix + "_" + str(genotypesCount))
countSpecificMetafileName = outTpedMetaFile.replace(outfile_prefix, outfile_prefix + "_" + str(genotypesCount))
with util.file.open_or_gzopen(countSpecificTpedName, 'a') as of1l, util.file.open_or_gzopen(countSpecificMetafileName, 'a') as of2l:
of1l.write(lineToWrite1)
of2l.write(lineToWrite2)
log.info("Genotype counts found: %s", str(list(recordLengths)))
def RunSnvplot(args):
cfg = Parse.generate_snvplot_cfg(args)
Parse.print_snvplot_options(cfg)
if not cfg['debug']:
logging.disable(logging.CRITICAL)
ro.r('suppressMessages(library(ggplot2))')
ro.r('suppressMessages(library(grid))')
handle=pysam.TabixFile(filename=cfg['file'],parser=pysam.asVCF())
header = [x for x in handle.header]
skip_rows = len(header)-1
cols = header[-1].split()
pcols = cfg['pcol'].split(',')
cols_extract = [cfg['chrcol'],cfg['bpcol']] + pcols
if cfg['qq_strat_freq']:
if cfg['freqcol'] not in cols:
print Process.Error("frequency column " + cfg['freqcol'] + " not found, unable to proceed with frequency stratified plots").out
return 1
else:
cols_extract = cols_extract + [cfg['freqcol']]
print "frequency column " + cfg['freqcol'] + " found"
if cfg['qq_strat_mac']:
if cfg['maccol'] not in cols:
print Process.Error("minor allele count column " + cfg['maccol'] + " not found, unable to proceed with minor allele count stratified plots").out
return 1
else:
cols_extract = cols_extract + [cfg['maccol']]
print "minor allele count column " + cfg['maccol'] + " found"
print "importing data"
r = pd.read_table(cfg['file'],sep='\t',skiprows=skip_rows,usecols=cols_extract,compression='gzip')
print str(r.shape[0]) + " total variants found"
for pcol in pcols:
print "plotting p-values for column " + pcol + " ..."
results = r[[cfg['chrcol'],cfg['bpcol'],cfg['freqcol'],pcol]] if cfg['freqcol'] in r else r[[cfg['chrcol'],cfg['bpcol'],pcol]]
results.dropna(inplace=True)
results = results[(results[pcol] > 0) & (results[pcol] <= 1)].reset_index(drop=True)
print " " + str(results.shape[0]) + " variants with plottable p-values"
results['logp'] = -1 * np.log10(results[pcol]) + 0.0
ro.globalenv['results'] = results
l = np.median(scipy.chi2.ppf([1-x for x in results[pcol].tolist()], df=1))/scipy.chi2.ppf(0.5,1)
# in R: median(qchisq(results$p, df=1, lower.tail=FALSE))/qchisq(0.5,1)
print " genomic inflation (all variants) = " + str(l)
if cfg['qq']:
print " generating standard qq plot"
print " minimum p-value: " + str(np.min(results[pcol]))
a = -1 * np.log10(ro.r('ppoints(' + str(len(results.index)) + ')'))
a.sort()
results.sort_values(by=['logp'], inplace=True)
print " maximum -1*log10(p-value): " + str(np.max(results['logp']))
ci_upper = -1 * np.log10(scipy.beta.ppf(0.95, range(1,len(results[pcol]) + 1), range(len(results[pcol]),0,-1)))
ci_upper.sort()
ci_lower = -1 * np.log10(scipy.beta.ppf(0.05, range(1,len(results[pcol]) + 1), range(len(results[pcol]),0,-1)))
ci_lower.sort()
ro.globalenv['df'] = ro.DataFrame({'a': ro.FloatVector(a), 'b': ro.FloatVector(results['logp']), 'ci_lower': ro.FloatVector(ci_lower), 'ci_upper': ro.FloatVector(ci_upper)})
dftext_label = 'lambda %~~% ' + str(l)
ro.globalenv['dftext'] = ro.DataFrame({'x': ro.r('Inf'), 'y': 0.5, 'lab': dftext_label})
if cfg['ext'] == 'tiff':
ggsave = 'ggsave(filename="%s",plot=pp,width=4,height=4,units="in",bg="white",compression="lzw",dpi=300)' % (cfg['out'] + '.' + pcol + '.qq.tiff')
elif cfg['ext'] == 'eps':
ggsave = 'ggsave(filename="%s",plot=pp,width=4,height=4,bg="white",horizontal=True)' % (cfg['out'] + '.' + pcol + '.qq.eps')
else:
ggsave = 'ggsave(filename="%s",plot=pp,width=4,height=4,bg="white")' % (cfg['out'] + '.' + pcol + '.qq.pdf')
ro.r("""
gp<-ggplot(df)
pp<-gp +
aes_string(x='a',y='b') +
geom_ribbon(aes_string(x='a',ymin='ci_lower',ymax='ci_upper'), data=df, alpha=0.25, fill='black') +
geom_point(size=2) +
geom_abline(intercept=0, slope=1, alpha=0.5) +
scale_x_discrete(expression(Expected~~-log[10](italic(p)))) +
scale_y_discrete(expression(Observed~~-log[10](italic(p)))) +
coord_fixed() +
theme_bw(base_size = 12) +
geom_text(aes_string(x='x', y='y', label='lab'), data = dftext, colour="black", vjust=0, hjust=1, size = 4, parse=TRUE) +
theme(axis.title.x = element_text(vjust=-0.5,size=14), axis.title.y = element_text(vjust=1,angle=90,size=14), legend.position = 'none',
panel.background = element_blank(), panel.border = element_blank(), panel.grid.minor = element_blank(),
panel.grid.major = element_blank(), axis.line = element_line(colour="black"), axis.text = element_text(size=12))
%s
""" % (ggsave))
if np.max(results['logp']) > cfg['crop']:
print " generating cropped standard qq plot"
ro.r('df$b[df$b > ' + str(cfg['crop']) + ']<-' + str(cfg['crop']))
ro.r('df$shape<-0')
ro.r('df$shape[df$b == ' + str(cfg['crop']) + ']<-1')
if cfg['ext'] == 'tiff':
ggsave = 'ggsave(filename="%s",plot=pp,width=4,height=4,units="in",bg="white",compression="lzw",dpi=300)' % (cfg['out'] + '.' + pcol + '.qq.cropped.tiff')
elif cfg['ext'] == 'eps':
ggsave = 'ggsave(filename="%s",plot=pp,width=4,height=4,bg="white",horizontal=True)' % (cfg['out'] + '.' + pcol + '.qq.cropped.eps')
else:
ggsave = 'ggsave(filename="%s",plot=pp,width=4,height=4,bg="white")' % (cfg['out'] + '.' + pcol + '.qq.cropped.pdf')
ro.r("""
gp<-ggplot(df)
pp<-gp +
aes_string(x='a',y='b') +
geom_ribbon(aes_string(x='a',ymin='ci_lower',ymax='ci_upper'), data=df, alpha=0.25, fill='black') +
geom_point(aes(shape=factor(shape)),size=2) +
geom_abline(intercept=0, slope=1, alpha=0.5) +
scale_x_discrete(expression(Expected~~-log[10](italic(p)))) +
scale_y_discrete(expression(Observed~~-log[10](italic(p)))) +
coord_fixed() +
theme_bw(base_size = 12) +
geom_text(aes_string(x='x', y='y', label='lab'), data = dftext, colour="black", vjust=0, hjust=1, size = 4, parse=TRUE) +
theme(axis.title.x = element_text(vjust=-0.5,size=14), axis.title.y = element_text(vjust=1,angle=90,size=14), legend.position = 'none',
panel.background = element_blank(), panel.border = element_blank(), panel.grid.minor = element_blank(),
panel.grid.major = element_blank(), axis.line = element_line(colour="black"), axis.text = element_text(size=12))
%s
""" % (ggsave))
if cfg['qq_strat_freq']:
print " generating frequency stratified qq plot"
strat_ticks = [0.005, 0.01, 0.03, 0.05]
results['UGA___QQ_BIN___'] = 'E'
results.loc[(results[cfg['freqcol']] >= 0.01) & (results[cfg['freqcol']] <= 0.99),'UGA___QQ_BIN___'] = 'D'
results.loc[(results[cfg['freqcol']] >= 0.03) & (results[cfg['freqcol']] <= 0.97),'UGA___QQ_BIN___'] = 'C'
results.loc[(results[cfg['freqcol']] >= 0.05) & (results[cfg['freqcol']] <= 0.95),'UGA___QQ_BIN___'] = 'B'
results.loc[(results[cfg['freqcol']] >= 0.1) & (results[cfg['freqcol']] <= 0.9),'UGA___QQ_BIN___'] = 'A'
lA='NA'
lB='NA'
lC='NA'
lD='NA'
lE='NA'
lE_n=len(results[pcol][(results[cfg['freqcol']] < 0.01) | (results[cfg['freqcol']] > 0.99)])
lD_n=len(results[pcol][((results[cfg['freqcol']] >= 0.01) & (results[cfg['freqcol']] < 0.03)) | ((results[cfg['freqcol']] <= 0.99) & (results[cfg['freqcol']] > 0.97))])
lC_n=len(results[pcol][((results[cfg['freqcol']] >= 0.03) & (results[cfg['freqcol']] < 0.05)) | ((results[cfg['freqcol']] <= 0.97) & (results[cfg['freqcol']] > 0.95))])
lB_n=len(results[pcol][((results[cfg['freqcol']] >= 0.05) & (results[cfg['freqcol']] < 0.1)) | ((results[cfg['freqcol']] <= 0.95) & (results[cfg['freqcol']] > 0.9))])
lA_n=len(results[pcol][(results[cfg['freqcol']] >= 0.1) & (results[cfg['freqcol']] <= 0.9)])
if lE_n > 0:
lE=np.median(scipy.chi2.ppf([1-x for x in results[pcol][(results[cfg['freqcol']] < 0.01) | (results[cfg['freqcol']] > 0.99)].tolist()], df=1))/scipy.chi2.ppf(0.5,1)
if lD_n > 0:
lD=np.median(scipy.chi2.ppf([1-x for x in results[pcol][((results[cfg['freqcol']] >= 0.01) & (results[cfg['freqcol']] < 0.03)) | ((results[cfg['freqcol']] <= 0.99) & (results[cfg['freqcol']] > 0.97))].tolist()], df=1))/scipy.chi2.ppf(0.5,1)
if lC_n > 0:
lC=np.median(scipy.chi2.ppf([1-x for x in results[pcol][((results[cfg['freqcol']] >= 0.03) & (results[cfg['freqcol']] < 0.05)) | ((results[cfg['freqcol']] <= 0.97) & (results[cfg['freqcol']] > 0.95))].tolist()], df=1))/scipy.chi2.ppf(0.5,1)
if lB_n > 0:
lB=np.median(scipy.chi2.ppf([1-x for x in results[pcol][((results[cfg['freqcol']] >= 0.05) & (results[cfg['freqcol']] < 0.1)) | ((results[cfg['freqcol']] <= 0.95) & (results[cfg['freqcol']] > 0.9))].tolist()], df=1))/scipy.chi2.ppf(0.5,1)
if lA_n > 0:
lA=np.median(scipy.chi2.ppf([1-x for x in results[pcol][(results[cfg['freqcol']] >= 0.1) & (results[cfg['freqcol']] <= 0.9)].tolist()], df=1))/scipy.chi2.ppf(0.5,1)
print " genomic inflation (MAF >= 10%, n=" + str(lA_n) + ") = " + str(lA)
print " genomic inflation (5% <= MAF < 10%, n=" + str(lB_n) + ") = " + str(lB)
print " genomic inflation (3% <= MAF < 5%, n=" + str(lC_n) + ") = " + str(lC)
print " genomic inflation (1% <= MAF < 3%, n=" + str(lD_n) + ") = " + str(lD)
print " genomic inflation (MAF < 1%, n=" + str(lE_n) + ") = " + str(lE)
a = np.array([])
b = np.array([])
c = np.array([])
results.sort_values(by=['logp'], inplace=True)
if len(results[results['UGA___QQ_BIN___'] == 'E'].index) > 0:
aa = -1 * np.log10(ro.r('ppoints(' + str(len(results[results['UGA___QQ_BIN___'] == 'E'].index)) + ')'))
aa.sort()
bb = results['logp'][results['UGA___QQ_BIN___'] == 'E']
#bb.sort()
cc = results['UGA___QQ_BIN___'][results['UGA___QQ_BIN___'] == 'E']
a = np.append(a,aa)
b = np.append(b,bb)
c = np.append(c,cc)
print " minimum p-value (MAF < 1%): " + str(np.min(results[pcol][results['UGA___QQ_BIN___'] == 'E']))
print " maximum -1*log10(p-value) (MAF < 1%): " + str(np.max(results['logp'][results['UGA___QQ_BIN___'] == 'E']))
if len(results[results['UGA___QQ_BIN___'] == 'D'].index) > 0:
aa = -1 * np.log10(ro.r('ppoints(' + str(len(results[results['UGA___QQ_BIN___'] == 'D'].index)) + ')'))
aa.sort()
bb = results['logp'][results['UGA___QQ_BIN___'] == 'D']
#bb.sort()
cc = results['UGA___QQ_BIN___'][results['UGA___QQ_BIN___'] == 'D']
a = np.append(a,aa)
b = np.append(b,bb)
c = np.append(c,cc)
print " minimum p-value (1% <= MAF < 3%): " + str(np.min(results[pcol][results['UGA___QQ_BIN___'] == 'D']))
print " maximum -1*log10(p-value) (1% <= MAF < 3%): " + str(np.max(results['logp'][results['UGA___QQ_BIN___'] == 'D']))
if len(results[results['UGA___QQ_BIN___'] == 'C'].index) > 0:
aa = -1 * np.log10(ro.r('ppoints(' + str(len(results[results['UGA___QQ_BIN___'] == 'C'].index)) + ')'))
aa.sort()
bb = results['logp'][results['UGA___QQ_BIN___'] == 'C']
#bb.sort()
cc = results['UGA___QQ_BIN___'][results['UGA___QQ_BIN___'] == 'C']
a = np.append(a,aa)
b = np.append(b,bb)
c = np.append(c,cc)
print " minimum p-value (3% <= MAF < 5%): " + str(np.min(results[pcol][results['UGA___QQ_BIN___'] == 'C']))
print " maximum -1*log10(p-value) (3% <= MAF < 5%): " + str(np.max(results['logp'][results['UGA___QQ_BIN___'] == 'C']))
if len(results[results['UGA___QQ_BIN___'] == 'B'].index) > 0:
aa = -1 * np.log10(ro.r('ppoints(' + str(len(results[results['UGA___QQ_BIN___'] == 'B'].index)) + ')'))
aa.sort()
bb = results['logp'][results['UGA___QQ_BIN___'] == 'B']
#bb.sort()
cc = results['UGA___QQ_BIN___'][results['UGA___QQ_BIN___'] == 'B']
a = np.append(a,aa)
b = np.append(b,bb)
c = np.append(c,cc)
print " minimum p-value (5% <= MAF < 10%): " + str(np.min(results[pcol][results['UGA___QQ_BIN___'] == 'B']))
print " maximum -1*log10(p-value) (5% <= MAF < 10%): " + str(np.max(results['logp'][results['UGA___QQ_BIN___'] == 'B']))
if len(results[results['UGA___QQ_BIN___'] == 'A'].index) > 0:
aa = -1 * np.log10(ro.r('ppoints(' + str(len(results[results['UGA___QQ_BIN___'] == 'A'].index)) + ')'))
aa.sort()
bb = results['logp'][results['UGA___QQ_BIN___'] == 'A']
#bb.sort()
cc = results['UGA___QQ_BIN___'][results['UGA___QQ_BIN___'] == 'A']
a = np.append(a,aa)
b = np.append(b,bb)
c = np.append(c,cc)
print " minimum p-value (MAF >= 10%): " + str(np.min(results[pcol][results['UGA___QQ_BIN___'] == 'A']))
print " maximum -1*log10(p-value) (MAF >= 10%): " + str(np.max(results['logp'][results['UGA___QQ_BIN___'] == 'A']))
ro.globalenv['df'] = ro.DataFrame({'a': ro.FloatVector(a), 'b': ro.FloatVector(b), 'UGA___QQ_BIN___': ro.StrVector(c)})
if cfg['ext'] == 'tiff':
ggsave = 'ggsave(filename="%s",plot=gp,width=4,height=4,units="in",bg="white",compression="lzw",dpi=300)' % (cfg['out'] + '.' + pcol + '.qq_strat_freq.tiff')
elif cfg['ext'] == 'eps':
ggsave = 'ggsave(filename="%s",plot=gp,width=4,height=4,bg="white",horizontal=True)' % (cfg['out'] + '.' + pcol + '.qq_strat_freq.eps')
else:
ggsave = 'ggsave(filename="%s",plot=gp,width=4,height=4,bg="white")' % (cfg['out'] + '.' + pcol + '.qq_strat_freq.pdf')
ro.r("""
gp<-ggplot(df, aes_string(x='a',y='b')) +
geom_point(aes_string(color='UGA___QQ_BIN___'), size=2) +
scale_colour_manual(values=c("E"="#a8ddb5", "D"="#7bccc4", "C"="#4eb3d3", "B"="#2b8cbe", "A"="#08589e"), labels=c("E"="MAF < 1%%","D"="1%% <= MAF < 3%%","C"="3%% <= MAF < 5%%","B"="5%% <= MAF < 10%%","A"="MAF >= 10%%")) +
geom_abline(intercept=0, slope=1, alpha=0.5) +
scale_x_discrete(expression(Expected~~-log[10](italic(p)))) +
scale_y_discrete(expression(Observed~~-log[10](italic(p)))) +
coord_fixed() +
theme_bw(base_size = 12) +
theme(axis.title.x = element_text(vjust=-0.5,size=14), axis.title.y = element_text(vjust=1,angle=90,size=14), legend.title = element_blank(),
legend.key.height = unit(0.1,"in"), legend.text = element_text(size=5), legend.key = element_blank(), legend.justification = c(0,1),
legend.position = c(0,1), panel.background = element_blank(), panel.border = element_blank(), panel.grid.minor = element_blank(),
panel.grid.major = element_blank(), axis.line = element_line(colour="black"), axis.text = element_text(size=12))
%s
""" % (ggsave))
if np.max(results['logp']) > cfg['crop']:
print " generating cropped frequency stratified qq plot"
ro.r('df$b[df$b > ' + str(cfg['crop']) + ']<-' + str(cfg['crop']))
ro.r('df$shape<-0')
ro.r('df$shape[df$b == ' + str(cfg['crop']) + ']<-1')
if cfg['ext'] == 'tiff':
ggsave = 'ggsave(filename="%s",plot=gp,width=4,height=4,units="in",bg="white",compression="lzw",dpi=300)' % (cfg['out'] + '.' + pcol + '.qq_strat_freq.cropped.tiff')
elif cfg['ext'] == 'eps':
ggsave = 'ggsave(filename="%s",plot=gp,width=4,height=4,bg="white",horizontal=True)' % (cfg['out'] + '.' + pcol + '.qq_strat_freq.cropped.eps')
else:
ggsave = 'ggsave(filename="%s",plot=gp,width=4,height=4,bg="white")' % (cfg['out'] + '.' + pcol + '.qq_strat_freq.cropped.pdf')
ro.r("""
gp<-ggplot(df, aes_string(x='a',y='b')) +
geom_point(aes(shape=factor(shape), color=UGA_MAF), size=2) +
scale_colour_manual(values=c("E"="#a8ddb5", "D"="#7bccc4", "C"="#4eb3d3", "B"="#2b8cbe", "A"="#08589e"), labels=c("E"="MAF < 1%%","D"="1%% <= MAF < 3%%","C"="3%% <= MAF < 5%%","B"="5%% <= MAF < 10%%","A"="MAF >= 10%%")) +
geom_abline(intercept=0, slope=1, alpha=0.5) +
scale_x_discrete(expression(Expected~~-log[10](italic(p)))) +
scale_y_discrete(expression(Observed~~-log[10](italic(p)))) +
coord_fixed() +
theme_bw(base_size = 12) +
guides(shape=FALSE) +
theme(axis.title.x = element_text(vjust=-0.5,size=14), axis.title.y = element_text(vjust=1,angle=90,size=14), legend.title = element_blank(),
legend.key.height = unit(0.1,"in"), legend.text = element_text(size=5), legend.key = element_blank(), legend.justification = c(0,1),
legend.position = c(0,1), panel.background = element_blank(), panel.border = element_blank(), panel.grid.minor = element_blank(),
panel.grid.major = element_blank(), axis.line = element_line(colour="black"), axis.text = element_text(size=12))
%s
""" % (ggsave))
#if cfg['qq_strat_mac']:
# print " generating minor allele count stratified qq plot"
#
# results['UGA_MAC'] = 'E'
# results.loc[results[cfg['maccol']] < 5),'UGA_MAC'] = 'D'
# results.loc[(results[cfg['maccol']] >= 0.03) & (results[cfg['maccol']] <= 0.97),'UGA_MAC'] = 'C'
# results.loc[(results[cfg['maccol']] >= 0.05) & (results[cfg['maccol']] <= 0.95),'UGA_MAC'] = 'B'
# results.loc[(results[cfg['maccol']] >= 0.1) & (results[cfg['maccol']] <= 0.9),'UGA_MAC'] = 'A'
# lA='NA'
# lB='NA'
# lC='NA'
# lD='NA'
# lE='NA'
# lE_n=len(results[pcol][(results[cfg['maccol']] < 0.01) | (results[cfg['maccol']] > 0.99)])
# lD_n=len(results[pcol][((results[cfg['maccol']] >= 0.01) & (results[cfg['maccol']] < 0.03)) | ((results[cfg['maccol']] <= 0.99) & (results[cfg['maccol']] > 0.97))])
# lC_n=len(results[pcol][((results[cfg['maccol']] >= 0.03) & (results[cfg['maccol']] < 0.05)) | ((results[cfg['maccol']] <= 0.97) & (results[cfg['maccol']] > 0.95))])
# lB_n=len(results[pcol][((results[cfg['maccol']] >= 0.05) & (results[cfg['maccol']] < 0.1)) | ((results[cfg['maccol']] <= 0.95) & (results[cfg['maccol']] > 0.9))])
# lA_n=len(results[pcol][(results[cfg['maccol']] >= 0.1) & (results[cfg['maccol']] <= 0.9)])
# if lE_n > 0:
# lE=np.median(scipy.chi2.ppf([1-x for x in results[pcol][(results[cfg['maccol']] < 0.01) | (results[cfg['maccol']] > 0.99)].tolist()], df=1))/scipy.chi2.ppf(0.5,1)
# if lD_n > 0:
# lD=np.median(scipy.chi2.ppf([1-x for x in results[pcol][((results[cfg['maccol']] >= 0.01) & (results[cfg['maccol']] < 0.03)) | ((results[cfg['maccol']] <= 0.99) & (results[cfg['maccol']] > 0.97))].tolist()], df=1))/scipy.chi2.ppf(0.5,1)
# if lC_n > 0:
# lC=np.median(scipy.chi2.ppf([1-x for x in results[pcol][((results[cfg['maccol']] >= 0.03) & (results[cfg['maccol']] < 0.05)) | ((results[cfg['maccol']] <= 0.97) & (results[cfg['maccol']] > 0.95))].tolist()], df=1))/scipy.chi2.ppf(0.5,1)
# if lB_n > 0:
# lB=np.median(scipy.chi2.ppf([1-x for x in results[pcol][((results[cfg['maccol']] >= 0.05) & (results[cfg['maccol']] < 0.1)) | ((results[cfg['maccol']] <= 0.95) & (results[cfg['maccol']] > 0.9))].tolist()], df=1))/scipy.chi2.ppf(0.5,1)
# if lA_n > 0:
# lA=np.median(scipy.chi2.ppf([1-x for x in results[pcol][(results[cfg['maccol']] >= 0.1) & (results[cfg['maccol']] <= 0.9)].tolist()], df=1))/scipy.chi2.ppf(0.5,1)
# print " genomic inflation (MAF >= 10%, n=" + str(lA_n) + ") = " + str(lA)
# print " genomic inflation (5% <= MAF < 10%, n=" + str(lB_n) + ") = " + str(lB)
# print " genomic inflation (3% <= MAF < 5%, n=" + str(lC_n) + ") = " + str(lC)
# print " genomic inflation (1% <= MAF < 3%, n=" + str(lD_n) + ") = " + str(lD)
# print " genomic inflation (MAF < 1%, n=" + str(lE_n) + ") = " + str(lE)
#
# a = np.array([])
# b = np.array([])
# c = np.array([])
# results.sort_values(by=['logp'], inplace=True)
# if len(results[results['UGA_MAC'] == 'E'].index) > 0:
# aa = -1 * np.log10(ro.r('ppoints(' + str(len(results[results['UGA_MAC'] == 'E'].index)) + ')'))
# aa.sort()
# bb = results['logp'][results['UGA_MAC'] == 'E']
# #bb.sort()
# cc = results['UGA_MAC'][results['UGA_MAC'] == 'E']
# a = np.append(a,aa)
# b = np.append(b,bb)
# c = np.append(c,cc)
# print " minimum p-value (MAF < 1%): " + str(np.min(results[pcol][results['UGA_MAC'] == 'E']))
# print " maximum -1*log10(p-value) (MAF < 1%): " + str(np.max(results['logp'][results['UGA_MAC'] == 'E']))
# if len(results[results['UGA_MAC'] == 'D'].index) > 0:
# aa = -1 * np.log10(ro.r('ppoints(' + str(len(results[results['UGA_MAC'] == 'D'].index)) + ')'))
# aa.sort()
# bb = results['logp'][results['UGA_MAC'] == 'D']
# #bb.sort()
# cc = results['UGA_MAC'][results['UGA_MAC'] == 'D']
# a = np.append(a,aa)
# b = np.append(b,bb)
# c = np.append(c,cc)
# print " minimum p-value (1% <= MAF < 3%): " + str(np.min(results[pcol][results['UGA_MAC'] == 'D']))
# print " maximum -1*log10(p-value) (1% <= MAF < 3%): " + str(np.max(results['logp'][results['UGA_MAC'] == 'D']))
# if len(results[results['UGA_MAC'] == 'C'].index) > 0:
# aa = -1 * np.log10(ro.r('ppoints(' + str(len(results[results['UGA_MAC'] == 'C'].index)) + ')'))
# aa.sort()
# bb = results['logp'][results['UGA_MAC'] == 'C']
# #bb.sort()
# cc = results['UGA_MAC'][results['UGA_MAC'] == 'C']
# a = np.append(a,aa)
# b = np.append(b,bb)
# c = np.append(c,cc)
# print " minimum p-value (3% <= MAF < 5%): " + str(np.min(results[pcol][results['UGA_MAC'] == 'C']))
# print " maximum -1*log10(p-value) (3% <= MAF < 5%): " + str(np.max(results['logp'][results['UGA_MAC'] == 'C']))
# if len(results[results['UGA_MAC'] == 'B'].index) > 0:
# aa = -1 * np.log10(ro.r('ppoints(' + str(len(results[results['UGA_MAC'] == 'B'].index)) + ')'))
# aa.sort()
# bb = results['logp'][results['UGA_MAC'] == 'B']
# #bb.sort()
# cc = results['UGA_MAC'][results['UGA_MAC'] == 'B']
# a = np.append(a,aa)
# b = np.append(b,bb)
# c = np.append(c,cc)
# print " minimum p-value (5% <= MAF < 10%): " + str(np.min(results[pcol][results['UGA_MAC'] == 'B']))
# print " maximum -1*log10(p-value) (5% <= MAF < 10%): " + str(np.max(results['logp'][results['UGA_MAC'] == 'B']))
# if len(results[results['UGA_MAC'] == 'A'].index) > 0:
# aa = -1 * np.log10(ro.r('ppoints(' + str(len(results[results['UGA_MAC'] == 'A'].index)) + ')'))
# aa.sort()
# bb = results['logp'][results['UGA_MAC'] == 'A']
# #bb.sort()
# cc = results['UGA_MAC'][results['UGA_MAC'] == 'A']
# a = np.append(a,aa)
# b = np.append(b,bb)
# c = np.append(c,cc)
# print " minimum p-value (MAF >= 10%): " + str(np.min(results[pcol][results['UGA_MAC'] == 'A']))
# print " maximum -1*log10(p-value) (MAF >= 10%): " + str(np.max(results['logp'][results['UGA_MAC'] == 'A']))
#
# ro.globalenv['df'] = ro.DataFrame({'a': ro.FloatVector(a), 'b': ro.FloatVector(b), 'UGA_MAC': ro.StrVector(c)})
#
# if cfg['ext'] == 'tiff':
# ggsave = 'ggsave(filename="%s",plot=gp,width=4,height=4,units="in",bg="white",compression="lzw",dpi=300)' % (cfg['out'] + '.' + pcol + '.qq_strat.tiff')
# elif cfg['ext'] == 'eps':
# ggsave = 'ggsave(filename="%s",plot=gp,width=4,height=4,bg="white",horizontal=True)' % (cfg['out'] + '.' + pcol + '.qq_strat.eps')
# else:
# ggsave = 'ggsave(filename="%s",plot=gp,width=4,height=4,bg="white")' % (cfg['out'] + '.' + pcol + '.qq_strat.pdf')
# ro.r("""
# gp<-ggplot(df, aes_string(x='a',y='b')) +
# geom_point(aes_string(color='UGA_MAC'), size=2) +
# scale_colour_manual(values=c("E"="#a8ddb5", "D"="#7bccc4", "C"="#4eb3d3", "B"="#2b8cbe", "A"="#08589e"), labels=c("E"="MAF < 1%%","D"="1%% <= MAF < 3%%","C"="3%% <= MAF < 5%%","B"="5%% <= MAF < 10%%","A"="MAF >= 10%%")) +
# geom_abline(intercept=0, slope=1, alpha=0.5) +
# scale_x_continuous(expression(Expected~~-log[10](italic(p)))) +
# scale_y_continuous(expression(Observed~~-log[10](italic(p)))) +
# theme_bw(base_size = 12) +
# theme(axis.title.x = element_text(vjust=-0.5,size=14), axis.title.y = element_text(vjust=1,angle=90,size=14), legend.title = element_blank(),
# legend.key.height = unit(0.1,"in"), legend.text = element_text(size=5), legend.key = element_blank(), legend.justification = c(0,1),
# legend.position = c(0,1), panel.background = element_blank(), panel.border = element_blank(), panel.grid.minor = element_blank(),
# panel.grid.major = element_blank(), axis.line = element_line(colour="black"), axis.text = element_text(size=12))
# %s
# """ % (ggsave))
#
# if np.max(results['logp']) > cfg['crop']:
# print " generating cropped frequency stratified qq plot"
# ro.r('df$b[df$b > ' + str(cfg['crop']) + ']<-' + str(cfg['crop']))
# ro.r('df$shape<-0')
# ro.r('df$shape[df$b == ' + str(cfg['crop']) + ']<-1')
# if cfg['ext'] == 'tiff':
# ggsave = 'ggsave(filename="%s",plot=gp,width=4,height=4,units="in",bg="white",compression="lzw",dpi=300)' % (cfg['out'] + '.' + pcol + '.qq_strat.cropped.tiff')
# elif cfg['ext'] == 'eps':
# ggsave = 'ggsave(filename="%s",plot=gp,width=4,height=4,bg="white",horizontal=True)' % (cfg['out'] + '.' + pcol + '.qq_strat.cropped.eps')
# else:
# ggsave = 'ggsave(filename="%s",plot=gp,width=4,height=4,bg="white")' % (cfg['out'] + '.' + pcol + '.qq_strat.cropped.pdf')
# ro.r("""
# gp<-ggplot(df, aes_string(x='a',y='b')) +
# geom_point(aes(shape=factor(shape), color=UGA_MAC), size=2) +
# scale_colour_manual(values=c("E"="#a8ddb5", "D"="#7bccc4", "C"="#4eb3d3", "B"="#2b8cbe", "A"="#08589e"), labels=c("E"="MAF < 1%%","D"="1%% <= MAF < 3%%","C"="3%% <= MAF < 5%%","B"="5%% <= MAF < 10%%","A"="MAF >= 10%%")) +
# geom_abline(intercept=0, slope=1, alpha=0.5) +
# scale_x_continuous(expression(Expected~~-log[10](italic(p)))) +
# scale_y_continuous(expression(Observed~~-log[10](italic(p)))) +
# theme_bw(base_size = 12) +
# guides(shape=FALSE) +
# theme(axis.title.x = element_text(vjust=-0.5,size=14), axis.title.y = element_text(vjust=1,angle=90,size=14), legend.title = element_blank(),
# legend.key.height = unit(0.1,"in"), legend.text = element_text(size=5), legend.key = element_blank(), legend.justification = c(0,1),
# legend.position = c(0,1), panel.background = element_blank(), panel.border = element_blank(), panel.grid.minor = element_blank(),
# panel.grid.major = element_blank(), axis.line = element_line(colour="black"), axis.text = element_text(size=12))
# %s
# """ % (ggsave))
if cfg['mht']:
print " generating standard manhattan plot"
print " minimum p-value: " + str(np.min(results[pcol]))
print " maximum -1*log10(p-value): " + str(np.max(results['logp']))
if cfg['gc'] and l > 1:
print " adjusting p-values for genomic inflation for p-value column " + pcol
results[pcol]=2 * scipy.norm.cdf(-1 * np.abs(scipy.norm.ppf(0.5*results[pcol]) / math.sqrt(l)))
print " minimum post-gc adjustment p-value: " + str(np.min(results[pcol]))
print " maximum post-gc adjustment -1*log10(p-value): " + str(np.max(results['logp']))
else:
print " skipping genomic inflation correction"
print " calculating genomic positions"
results.sort_values(by=[cfg['chrcol'],cfg['bpcol']], inplace=True)
ticks = []
lastbase = 0
results['gpos'] = 0
nchr = len(list(np.unique(results[cfg['chrcol']].values)))
chrs = np.unique(results[cfg['chrcol']].values)
if cfg['color']:
colours = ["#08306B","#41AB5D","#000000","#F16913","#3F007D","#EF3B2C","#08519C","#238B45","#252525","#D94801","#54278F","#CB181D","#2171B5","#006D2C","#525252","#A63603","#6A51A3","#A50F15","#4292C6","#00441B","#737373","#7F2704","#807DBA","#67000D"]
else:
colours = ["#08589e","#4eb3d3","#08589e","#4eb3d3","#08589e","#4eb3d3","#08589e","#4eb3d3","#08589e","#4eb3d3","#08589e","#4eb3d3","#08589e","#4eb3d3","#08589e","#4eb3d3","#08589e","#4eb3d3","#08589e","#4eb3d3","#08589e","#4eb3d3","#08589e","#4eb3d3"]
if nchr == 1:
results['gpos'] = results[cfg['bpcol']]
results['colours'] = "#08589e"
if results['gpos'].max() - results['gpos'].min() <= 1000:
ticks = [x for x in range(results['gpos'].min(),results['gpos'].max()) if x % 100 == 0]
elif results['gpos'].max() - results['gpos'].min() <= 10000:
ticks = [x for x in range(results['gpos'].min(),results['gpos'].max()) if x % 1000 == 0]
elif results['gpos'].max() - results['gpos'].min() <= 100000:
ticks = [x for x in range(results['gpos'].min(),results['gpos'].max()) if x % 10000 == 0]
elif results['gpos'].max() - results['gpos'].min() <= 200000:
ticks = [x for x in range(results['gpos'].min(),results['gpos'].max()) if x % 20000 == 0]
elif results['gpos'].max() - results['gpos'].min() <= 300000:
ticks = [x for x in range(results['gpos'].min(),results['gpos'].max()) if x % 30000 == 0]
elif results['gpos'].max() - results['gpos'].min() <= 400000:
ticks = [x for x in range(results['gpos'].min(),results['gpos'].max()) if x % 40000 == 0]
elif results['gpos'].max() - results['gpos'].min() <= 500000:
ticks = [x for x in range(results['gpos'].min(),results['gpos'].max()) if x % 50000 == 0]
elif results['gpos'].max() - results['gpos'].min() <= 600000:
ticks = [x for x in range(results['gpos'].min(),results['gpos'].max()) if x % 60000 == 0]
elif results['gpos'].max() - results['gpos'].min() <= 700000:
ticks = [x for x in range(results['gpos'].min(),results['gpos'].max()) if x % 70000 == 0]
elif results['gpos'].max() - results['gpos'].min() <= 800000:
ticks = [x for x in range(results['gpos'].min(),results['gpos'].max()) if x % 80000 == 0]
elif results['gpos'].max() - results['gpos'].min() <= 900000:
ticks = [x for x in range(results['gpos'].min(),results['gpos'].max()) if x % 90000 == 0]
elif results['gpos'].max() - results['gpos'].min() <= 1000000:
ticks = [x for x in range(results['gpos'].min(),results['gpos'].max()) if x % 100000 == 0]
elif results['gpos'].max() - results['gpos'].min() <= 10000000:
ticks = [x for x in range(results['gpos'].min(),results['gpos'].max()) if x % 1000000 == 0]
elif results['gpos'].max() - results['gpos'].min() <= 100000000:
ticks = [x for x in range(results['gpos'].min(),results['gpos'].max()) if x % 10000000 == 0]
elif results['gpos'].max() - results['gpos'].min() > 100000000:
ticks = [x for x in range(results['gpos'].min(),results['gpos'].max()) if x % 25000000 == 0]
else:
results['colours'] = "#000000"
for i in range(len(chrs)):
print " processed chromosome " + str(int(chrs[i]))
if i == 0:
results.loc[results[cfg['chrcol']] == chrs[i],'gpos'] = results.loc[results[cfg['chrcol']] == chrs[i],cfg['bpcol']]
else:
lastbase = lastbase + results.loc[results[cfg['chrcol']] == chrs[i-1],cfg['bpcol']].iloc[-1]
results.loc[results[cfg['chrcol']] == chrs[i],'gpos'] = (results.loc[results[cfg['chrcol']] == chrs[i],cfg['bpcol']]) + lastbase
if results.loc[results[cfg['chrcol']] == chrs[i]].shape[0] > 1:
ticks.append(results.loc[results[cfg['chrcol']] == chrs[i],'gpos'].iloc[0] + (results.loc[results[cfg['chrcol']] == chrs[i],'gpos'].iloc[-1] - results.loc[results[cfg['chrcol']] == chrs[i],'gpos'].iloc[0])/2)
else:
ticks.append(results.loc[results[cfg['chrcol']] == chrs[i],'gpos'].iloc[0])
results.loc[results[cfg['chrcol']] == chrs[i],'colours'] = colours[int(chrs[i])]
results['logp'] = -1 * np.log10(results[pcol])
if results.shape[0] >= 1000000:
sig = 5.4e-8
else:
sig = 0.05 / results.shape[0]
print " significance level set to p-value = " + str(sig) + " (-1*log10(p-value) = " + str(-1 * np.log10(sig)) + ")"
print " " + str(len(results[pcol][results[pcol] <= sig])) + " genome wide significant variants"
chr = results[cfg['chrcol']][0]
maxy=int(max(np.ceil(-1 * np.log10(sig)),np.ceil(results['logp'].max())))
if maxy > 20:
y_breaks = range(0,maxy,5)
y_labels = range(0,maxy,5)
else:
y_breaks = range(0,maxy)
y_labels = range(0,maxy)
ro.globalenv['df'] = ro.DataFrame({'gpos': ro.FloatVector(results['gpos']), 'logp': ro.FloatVector(results['logp']), 'colours': ro.FactorVector(results['colours'])})
ro.globalenv['ticks'] = ro.FloatVector(ticks)
ro.globalenv['labels'] = ro.Vector(["{:,}".format(x/1000) for x in ticks])
ro.globalenv['colours'] = ro.StrVector(colours)
ro.globalenv['chrs'] = ro.FloatVector(chrs)
print " generating manhattan plot"
if cfg['ext'] == 'tiff':
ggsave = 'ggsave(filename="%s",plot=gp,width=16,height=4,units="in",bg="white",compression="lzw",dpi=300)' % (cfg['out'] + '.' + pcol + '.mht.tiff')
elif cfg['ext'] == 'eps':
ggsave = 'ggsave(filename="%s",plot=gp,width=16,height=4,bg="white",horizontal=True)' % (cfg['out'] + '.' + pcol + '.mht.eps')
else:
ggsave = 'ggsave(filename="%s",plot=gp,width=16,height=4,bg="white")' % (cfg['out'] + '.' + pcol + '.mht.pdf')
if nchr == 1:
ro.r("""
gp<-ggplot(df, aes_string(x='gpos',y='logp')) +
geom_hline(yintercept = -1 * log10(%g),colour="#B8860B", linetype=5, size = 0.25) +
geom_point(size=1.5) +
scale_x_continuous(expression(Chromosome~~%d~~(kb))'),breaks=ticks,labels=labels) + \
scale_y_continuous(expression(-log[10](italic(p))),breaks=seq(0,%d,1),limits=c(0,%d)) + \
theme_bw(base_size = 8) + \
theme(axis.title.x = element_text(vjust=-0.5,size=14), axis.title.y = element_text(vjust=1,angle=90,size=14),
panel.background = element_blank(), panel.border = element_blank(), panel.grid.minor = element_blank(),
panel.grid.major = element_blank(), axis.line = element_line(colour="black"), axis.title = element_text(size=10),
axis.text = element_text(size=12), legend.position = 'none')
%s
""" % (sig, chr, maxy, maxy, ggsave))
else:
ro.r("""
gp = ggplot(df, aes_string(x='gpos',y='logp',colour='colours')) +
geom_hline(yintercept = -1 * log10(%g),colour="#B8860B", linetype=5, size = 0.25) +
geom_point(size=1.5) +
scale_colour_manual(values=colours) +
scale_x_continuous(expression(Chromosome),breaks=ticks,labels=chrs) +
scale_y_continuous(expression(-log[10](italic(p))),breaks=seq(0,%d,1),limits=c(0,%d)) +
theme_bw(base_size = 8) +
theme(axis.title.x = element_text(vjust=-0.5,size=14), axis.title.y = element_text(vjust=1,angle=90,size=14),
panel.background = element_blank(), panel.border = element_blank(), panel.grid.minor = element_blank(),
panel.grid.major = element_blank(), axis.line = element_line(colour="black"), axis.title = element_text(size=10),
axis.text = element_text(size=12), legend.position = 'none')
%s
""" % (sig, maxy, maxy, ggsave))
if maxy > cfg['crop']:
maxy = cfg['crop']
ro.r('df$logp[df$logp > ' + str(cfg['crop']) + ']<-' + str(cfg['crop']))
ro.r('df$shape<-0')
ro.r('df$shape[df$logp == ' + str(cfg['crop']) + ']<-1')
print " generating cropped manhattan plot"
if cfg['ext'] == 'tiff':
ggsave = 'ggsave(filename="%s",plot=gp,width=16,height=4,units="in",bg="white",compression="lzw",dpi=300)' % (cfg['out'] + '.' + pcol + '.mht.cropped.tiff')
elif cfg['ext'] == 'eps':
ggsave = 'ggsave(filename="%s",plot=gp,width=16,height=4,bg="white",horizontal=True)' % (cfg['out'] + '.' + pcol + '.mht.cropped.eps')
else:
ggsave = 'ggsave(filename="%s",plot=gp,width=16,height=4,bg="white")' % (cfg['out'] + '.' + pcol + '.mht.cropped.pdf')
if nchr == 1:
ro.r("""
gp<-ggplot(df, aes_string(x='gpos',y='logp')) +
geom_hline(yintercept = -1 * log10(%g),colour="#B8860B", linetype=5, size = 0.25) +
geom_point(aes(shape=factor(shape)),size=1.5) +
scale_x_continuous(expression(Chromosome~~%d~~(kb))'),breaks=ticks,labels=labels) +
scale_y_continuous(expression(-log[10](italic(p))),breaks=seq(0,%d,1),limits=c(0,%d)) +
theme_bw(base_size = 8) +
theme(axis.title.x = element_text(vjust=-0.5,size=14), axis.title.y = element_text(vjust=1,angle=90,size=14),
panel.background = element_blank(), panel.border = element_blank(), panel.grid.minor = element_blank(),
panel.grid.major = element_blank(), axis.line = element_line(colour="black"), axis.title = element_text(size=10),
axis.text = element_text(size=12), legend.position = 'none')
%s
""" % (sig, chr, maxy, maxy, ggsave))
else:
ro.r("""
gp = ggplot(df, aes_string(x='gpos',y='logp',colour='colours')) +
geom_hline(yintercept = -1 * log10(%g),colour="#B8860B", linetype=5, size = 0.25) +
geom_point(aes(shape=factor(shape)),size=1.5) +
scale_colour_manual(values=colours) +
scale_x_continuous(expression(Chromosome),breaks=ticks,labels=chrs) +
scale_y_continuous(expression(-log[10](italic(p))),breaks=seq(0,%d,1),limits=c(0,%d)) +
theme_bw(base_size = 8) +
theme(axis.title.x = element_text(vjust=-0.5,size=14), axis.title.y = element_text(vjust=1,angle=90,size=14),
panel.background = element_blank(), panel.border = element_blank(), panel.grid.minor = element_blank(),
panel.grid.major = element_blank(), axis.line = element_line(colour="black"), axis.title = element_text(size=8),
axis.text = element_text(size=12), legend.position = 'none')
%s
""" % (sig, maxy, maxy, ggsave))
print "process complete"
return 0
def __init__(self, file_path, parser=pysam.asVCF()):
self.vcf_file_path = file_path
self.tabix_file = pysam.TabixFile(file_path, parser=parser)
self.sample_names = self.read_sample_names()
self.clens = self.contig_lengths()
self.indexDelta = -1 if tuple(map(int, pysam.__version__.split('.'))) > (0,5) else 0
def RunSnvplot(args):
cfg = Parse.generate_snvplot_cfg(args)
Parse.print_snvplot_options(cfg)
if not cfg['debug']:
logging.disable(logging.CRITICAL)
ro.r('suppressMessages(library(ggplot2))')
ro.r('suppressMessages(library(grid))')
ro.r('suppressMessages(library(RColorBrewer))')
handle=pysam.TabixFile(filename=cfg['file'],parser=pysam.asVCF())
header = [x for x in handle.header]
skip_rows = len(header)-1
cols = header[-1].split()
pcols = cfg['pcol'].split(',')
cols_extract = [cfg['chrcol'],cfg['bpcol']] + pcols
if cfg['qq_strat_freq']:
if cfg['freqcol'] not in cols:
print Process.Error("frequency column " + cfg['freqcol'] + " not found, unable to proceed with frequency stratified plots").out
return 1
else:
cols_extract = cols_extract + [cfg['freqcol']]
print "frequency column " + cfg['freqcol'] + " found"
if cfg['qq_strat_mac']:
if cfg['maccol'] not in cols:
print Process.Error("minor allele count column " + cfg['maccol'] + " not found, unable to proceed with minor allele count stratified plots").out
return 1
else:
cols_extract = cols_extract + [cfg['maccol']]
print "minor allele count column " + cfg['maccol'] + " found"
print "importing data"
r = pd.read_table(cfg['file'],sep='\t',skiprows=skip_rows,usecols=cols_extract,compression='gzip')
print str(r.shape[0]) + " total variants found"
for pcol in pcols:
print "plotting p-values for column " + pcol + " ..."
extract_cols = [cfg['chrcol'],cfg['bpcol'],pcol]
if cfg['freqcol'] in r:
extract_cols = extract_cols + [cfg['freqcol']]
if cfg['maccol'] in r:
extract_cols = extract_cols + [cfg['maccol']]
results = r[extract_cols]
results.dropna(inplace=True)
results = results[(results[pcol] > 0) & (results[pcol] <= 1)].reset_index(drop=True)
print " " + str(results.shape[0]) + " variants with plottable p-values"
results['logp'] = -1 * np.log10(results[pcol]) + 0.0
ro.globalenv['results'] = results
l = np.median(scipy.chi2.ppf([1-x for x in results[pcol].tolist()], df=1))/scipy.chi2.ppf(0.5,1)
# in R: median(qchisq(results$p, df=1, lower.tail=FALSE))/qchisq(0.5,1)
print " genomic inflation (all variants) = " + str(l)
if cfg['qq']:
print " generating standard qq plot"
print " minimum p-value: " + str(np.min(results[pcol]))
a = -1 * np.log10(ro.r('ppoints(' + str(len(results.index)) + ')'))
a.sort()
results.sort_values(by=['logp'], inplace=True)
print " maximum -1*log10(p-value): " + str(np.max(results['logp']))
ci_upper = -1 * np.log10(scipy.beta.ppf(0.95, range(1,len(results[pcol]) + 1), range(len(results[pcol]),0,-1)))
ci_upper.sort()
ci_lower = -1 * np.log10(scipy.beta.ppf(0.05, range(1,len(results[pcol]) + 1), range(len(results[pcol]),0,-1)))
ci_lower.sort()
ro.globalenv['df'] = ro.DataFrame({'a': ro.FloatVector(a), 'b': ro.FloatVector(results['logp']), 'ci_lower': ro.FloatVector(ci_lower), 'ci_upper': ro.FloatVector(ci_upper)})
dftext_label = 'lambda %~~% ' + str(round(l,3))
ro.globalenv['dftext'] = ro.DataFrame({'x': ro.r('Inf'), 'y': 0.5, 'lab': dftext_label})
if cfg['ext'] == 'tiff':
ggsave = 'ggsave(filename="%s",plot=pp,width=4,height=4,units="in",bg="white",compression="lzw",dpi=300)' % (cfg['out'] + '.' + pcol + '.qq.tiff')
elif cfg['ext'] == 'png':
ggsave = 'ggsave(filename="%s",plot=pp,width=4,height=4,units="in",bg="white",dpi=300)' % (cfg['out'] + '.' + pcol + '.qq.png')
elif cfg['ext'] == 'eps':
ggsave = 'ggsave(filename="%s",plot=pp,width=4,height=4,bg="white")' % (cfg['out'] + '.' + pcol + '.qq.eps')
else:
ggsave = 'ggsave(filename="%s",plot=pp,width=4,height=4,bg="white")' % (cfg['out'] + '.' + pcol + '.qq.pdf')
ro.r("""
gp<-ggplot(df)
pp<-gp +
aes_string(x='a',y='b') +
geom_ribbon(aes_string(x='a',ymin='ci_lower',ymax='ci_upper'), data=df, alpha=0.25, fill='black') +
geom_point(size=2) +
geom_abline(intercept=0, slope=1, alpha=0.5) +
scale_x_discrete(expression(Expected~~-log[10](italic(p)))) +
scale_y_discrete(expression(Observed~~-log[10](italic(p)))) +
coord_fixed() +
theme_bw(base_size = 12) +
geom_text(aes_string(x='x', y='y', label='lab'), data = dftext, colour="black", vjust=0, hjust=1, size = 4, parse=TRUE) +
theme(axis.title.x = element_text(vjust=-0.5,size=14), axis.title.y = element_text(vjust=1,angle=90,size=14), legend.position = 'none',
panel.background = element_blank(), panel.border = element_blank(), panel.grid.minor = element_blank(),
panel.grid.major = element_blank(), axis.line = element_line(colour="black"), axis.text = element_text(size=12))
%s
""" % (ggsave))
if np.max(results['logp']) > cfg['crop']:
print " generating cropped standard qq plot"
ro.r('df$b[df$b > ' + str(cfg['crop']) + ']<-' + str(cfg['crop']))
ro.r('df$shape<-0')
ro.r('df$shape[df$b == ' + str(cfg['crop']) + ']<-1')
if cfg['ext'] == 'tiff':
ggsave = 'ggsave(filename="%s",plot=pp,width=4,height=4,units="in",bg="white",compression="lzw",dpi=300)' % (cfg['out'] + '.' + pcol + '.qq.cropped.tiff')
elif cfg['ext'] == 'png':
ggsave = 'ggsave(filename="%s",plot=pp,width=4,height=4,units="in",bg="white",dpi=300)' % (cfg['out'] + '.' + pcol + '.qq.cropped.png')
elif cfg['ext'] == 'eps':
ggsave = 'ggsave(filename="%s",plot=pp,width=4,height=4,bg="white")' % (cfg['out'] + '.' + pcol + '.qq.cropped.eps')
else:
ggsave = 'ggsave(filename="%s",plot=pp,width=4,height=4,bg="white")' % (cfg['out'] + '.' + pcol + '.qq.cropped.pdf')
ro.r("""
gp<-ggplot(df)
pp<-gp +
aes_string(x='a',y='b') +
geom_ribbon(aes_string(x='a',ymin='ci_lower',ymax='ci_upper'), data=df, alpha=0.25, fill='black') +
geom_point(aes(shape=factor(shape)),size=2) +
geom_abline(intercept=0, slope=1, alpha=0.5) +
scale_x_discrete(expression(Expected~~-log[10](italic(p)))) +
scale_y_discrete(expression(Observed~~-log[10](italic(p)))) +
coord_fixed() +
theme_bw(base_size = 12) +
geom_text(aes_string(x='x', y='y', label='lab'), data = dftext, colour="black", vjust=0, hjust=1, size = 4, parse=TRUE) +
theme(axis.title.x = element_text(vjust=-0.5,size=14), axis.title.y = element_text(vjust=1,angle=90,size=14), legend.position = 'none',
panel.background = element_blank(), panel.border = element_blank(), panel.grid.minor = element_blank(),
panel.grid.major = element_blank(), axis.line = element_line(colour="black"), axis.text = element_text(size=12))
%s
""" % (ggsave))
def ppoints(n, a):
try:
n = np.float(len(n))
except TypeError:
n = np.float(n)
return (np.arange(n) + 1 - a)/(n + 1 - 2*a)
if cfg['qq_strat_freq']:
print " generating frequency stratified qq plot"
strat_ticks = np.sort([np.float(x) for x in cfg['freq_ticks'].split(',')])
results['UGA___QQ_BIN___'] = 0
for i in xrange(len(strat_ticks)):
results.loc[(results[cfg['freqcol']] >= strat_ticks[i]) & (results[cfg['freqcol']] <= 1-strat_ticks[i]),'UGA___QQ_BIN___'] = i+1
bin_values = results['UGA___QQ_BIN___'].value_counts()
for i in xrange(len(strat_ticks)+1):
if i not in bin_values.index:
bin_values[i] = 0
counts = pd.DataFrame(bin_values)
counts['lambda'] = np.nan
results['description'] = 'NA'
for i in xrange(len(strat_ticks)+1):
if counts.loc[i,'UGA___QQ_BIN___'] > 0:
counts.loc[i,'lambda'] = np.median(scipy.chi2.ppf([1-x for x in results[pcol][results['UGA___QQ_BIN___'] == i].tolist()], df=1))/scipy.chi2.ppf(0.5,1)
else:
counts.loc[i,'lambda'] = np.nan
if i == 0:
results.loc[results['UGA___QQ_BIN___'] == i,'description'] = "(0," + str(strat_ticks[i]) + ") ~" + str(round(counts.loc[i,'lambda'],3))
print " MAF (0," + str(strat_ticks[i]) + "): n=" + str(np.int(counts.loc[i,'UGA___QQ_BIN___'])) + ", lambda=" + str(counts.loc[i,'lambda'])
elif i < len(strat_ticks):
results.loc[results['UGA___QQ_BIN___'] == i,'description'] = "[" + str(strat_ticks[i-1]) + "," + str(strat_ticks[i]) + ") ~" + str(round(counts.loc[i,'lambda'],3))
print " MAF [" + str(strat_ticks[i-1]) + "," + str(strat_ticks[i]) + "): n=" + str(np.int(counts.loc[i,'UGA___QQ_BIN___'])) + ", lambda=" + str(counts.loc[i,'lambda'])
else:
results.loc[results['UGA___QQ_BIN___'] == i,'description'] = "[" + str(strat_ticks[i-1]) + ",0.5] ~" + str(round(counts.loc[i,'lambda'],3))
print " MAF [" + str(strat_ticks[i-1]) + ",0.5]: n=" + str(np.int(counts.loc[i,'UGA___QQ_BIN___'])) + ", lambda=" + str(counts.loc[i,'lambda'])
results.sort_values(['UGA___QQ_BIN___','logp'],inplace=True)
results['expected'] = 0
for i in counts.index:
if counts.loc[i,'UGA___QQ_BIN___'] > 0:
results.loc[results['UGA___QQ_BIN___'] == i,'expected'] = np.sort(-1 * np.log10(ppoints(len(results.loc[results['UGA___QQ_BIN___'] == i,'expected']),0)))
ro.globalenv['df'] = ro.DataFrame({'expected': ro.FloatVector(results['expected']), 'logp': ro.FloatVector(results['logp']), 'UGA___QQ_BIN___': ro.IntVector(results['UGA___QQ_BIN___']), 'description': ro.StrVector(results['description'])})
ro.r("df<-df[order(df$UGA___QQ_BIN___),]")
ro.r("df$description<-ordered(df$description,levels=unique(df$description))")
if cfg['ext'] == 'tiff':
ggsave = 'ggsave(filename="%s",plot=gp,width=4,height=4,units="in",bg="white",compression="lzw",dpi=300)' % (cfg['out'] + '.' + pcol + '.qq_strat_freq.tiff')
elif cfg['ext'] == 'png':
ggsave = 'ggsave(filename="%s",plot=gp,width=4,height=4,units="in",bg="white",dpi=300)' % (cfg['out'] + '.' + pcol + '.qq_strat_freq.png')
elif cfg['ext'] == 'eps':
ggsave = 'ggsave(filename="%s",plot=gp,width=4,height=4,bg="white")' % (cfg['out'] + '.' + pcol + '.qq_strat_freq.eps')
else:
ggsave = 'ggsave(filename="%s",plot=gp,width=4,height=4,bg="white")' % (cfg['out'] + '.' + pcol + '.qq_strat_freq.pdf')
ro.r("""
gp<-ggplot(df, aes_string(x='expected',y='logp')) +
geom_point(aes_string(color='description'), size=2) +
scale_colour_manual(values=colorRampPalette(brewer.pal(9,"Blues"))(length(unique(df$description))+2)[3:(length(unique(df$description))+2)]) +
geom_abline(intercept=0, slope=1, alpha=0.5) +
scale_x_discrete(expression(Expected~~-log[10](italic(p)))) +
scale_y_discrete(expression(Observed~~-log[10](italic(p)))) +
coord_fixed() +
theme_bw(base_size = 12) +
theme(axis.title.x = element_text(vjust=-0.5,size=14), axis.title.y = element_text(vjust=1,angle=90,size=14), legend.title = element_blank(),
legend.key.height = unit(0.1,"in"), legend.text = element_text(size=6), legend.key = element_blank(), legend.justification = c(0,1),
legend.position = c(0,1), panel.background = element_blank(), panel.border = element_blank(), panel.grid.minor = element_blank(),
panel.grid.major = element_blank(), axis.line = element_line(colour="black"), axis.text = element_text(size=12))
%s
""" % (ggsave))
if np.max(results['logp']) > cfg['crop']:
print " generating cropped frequency stratified qq plot"
ro.r('df$logp[df$logp > ' + str(cfg['crop']) + ']<-' + str(cfg['crop']))
ro.r('df$shape<-0')
ro.r('df$shape[df$logp == ' + str(cfg['crop']) + ']<-1')
if cfg['ext'] == 'tiff':
ggsave = 'ggsave(filename="%s",plot=gp,width=4,height=4,units="in",bg="white",compression="lzw",dpi=300)' % (cfg['out'] + '.' + pcol + '.qq_strat_freq.cropped.tiff')
elif cfg['ext'] == 'png':
ggsave = 'ggsave(filename="%s",plot=gp,width=4,height=4,units="in",bg="white",dpi=300)' % (cfg['out'] + '.' + pcol + '.qq_strat_freq.cropped.png')
elif cfg['ext'] == 'eps':
ggsave = 'ggsave(filename="%s",plot=gp,width=4,height=4,bg="white")' % (cfg['out'] + '.' + pcol + '.qq_strat_freq.cropped.eps')
else:
ggsave = 'ggsave(filename="%s",plot=gp,width=4,height=4,bg="white")' % (cfg['out'] + '.' + pcol + '.qq_strat_freq.cropped.pdf')
ro.r("""
gp<-ggplot(df, aes_string(x='expected',y='logp')) +
geom_point(aes(shape=factor(shape), color=description), size=2) +
scale_colour_manual(values=colorRampPalette(brewer.pal(9,"Blues"))(length(unique(df$description))+2)[3:(length(unique(df$description))+2)]) +
geom_abline(intercept=0, slope=1, alpha=0.5) +
scale_x_discrete(expression(Expected~~-log[10](italic(p)))) +
scale_y_discrete(expression(Observed~~-log[10](italic(p)))) +
coord_fixed() +
theme_bw(base_size = 12) +
guides(shape=FALSE) +
theme(axis.title.x = element_text(vjust=-0.5,size=14), axis.title.y = element_text(vjust=1,angle=90,size=14), legend.title = element_blank(),
legend.key.height = unit(0.1,"in"), legend.text = element_text(size=6), legend.key = element_blank(), legend.justification = c(0,1),
legend.position = c(0,1), panel.background = element_blank(), panel.border = element_blank(), panel.grid.minor = element_blank(),
panel.grid.major = element_blank(), axis.line = element_line(colour="black"), axis.text = element_text(size=12))
%s
""" % (ggsave))
if cfg['qq_strat_mac']:
print " generating minor allele count stratified qq plot"
strat_ticks = np.sort([np.float(x) for x in cfg['mac_ticks'].split(',')])
results['UGA___QQ_BIN___'] = 0
for i in xrange(len(strat_ticks)):
results.loc[results[cfg['maccol']] >= strat_ticks[i],'UGA___QQ_BIN___'] = i+1
bin_values = results['UGA___QQ_BIN___'].value_counts()
for i in xrange(len(strat_ticks)+1):
if i not in bin_values.index:
bin_values[i] = 0
counts = pd.DataFrame(bin_values)
counts['lambda'] = 0
results['description'] = 'NA'
for i in np.sort(counts.index):
if counts.loc[i,'UGA___QQ_BIN___'] > 0:
counts.loc[i,'lambda'] = np.median(scipy.chi2.ppf([1-x for x in results[pcol][results['UGA___QQ_BIN___'] == i].tolist()], df=1))/scipy.chi2.ppf(0.5,1)
else:
counts.loc[i,'lambda'] = np.nan
if i == 0:
results.loc[results['UGA___QQ_BIN___'] == i,'description'] = "(0," + str(int(strat_ticks[i])) + ") ~" + str(round(counts.loc[i,'lambda'],3))
print " MAC (0," + str(int(strat_ticks[i])) + "): n=" + str(np.int(counts.loc[i,'UGA___QQ_BIN___'])) + ", lambda=" + str(counts.loc[i,'lambda'])
elif i < len(strat_ticks):
results.loc[results['UGA___QQ_BIN___'] == i,'description'] = "[" + str(int(strat_ticks[i-1])) + "," + str(int(strat_ticks[i])) + ") ~" + str(round(counts.loc[i,'lambda'],3))
print " MAC [" + str(int(strat_ticks[i-1])) + "," + str(int(strat_ticks[i])) + "): n=" + str(np.int(counts.loc[i,'UGA___QQ_BIN___'])) + ", lambda=" + str(counts.loc[i,'lambda'])
else:
results.loc[results['UGA___QQ_BIN___'] == i,'description'] = "[" + str(int(strat_ticks[i-1])) + ",...] ~" + str(round(counts.loc[i,'lambda'],3))
print " MAC [" + str(int(strat_ticks[i-1])) + ",...]: n=" + str(np.int(counts.loc[i,'UGA___QQ_BIN___'])) + ", lambda=" + str(counts.loc[i,'lambda'])
results.sort_values(['UGA___QQ_BIN___','logp'],inplace=True)
results['expected'] = 0
for i in counts.index:
results.loc[results['UGA___QQ_BIN___'] == i,'expected'] = np.sort(-1 * np.log10(ppoints(len(results.loc[results['UGA___QQ_BIN___'] == i,'expected']),0)))
ro.globalenv['df'] = ro.DataFrame({'expected': ro.FloatVector(results['expected']), 'logp': ro.FloatVector(results['logp']), 'UGA___QQ_BIN___': ro.IntVector(results['UGA___QQ_BIN___']), 'description': ro.StrVector(results['description'])})
ro.r("df<-df[order(df$UGA___QQ_BIN___),]")
ro.r("df$description<-ordered(df$description,levels=unique(df$description))")
if cfg['ext'] == 'tiff':
ggsave = 'ggsave(filename="%s",plot=gp,width=4,height=4,units="in",bg="white",compression="lzw",dpi=300)' % (cfg['out'] + '.' + pcol + '.qq_strat_mac.tiff')
elif cfg['ext'] == 'png':
ggsave = 'ggsave(filename="%s",plot=gp,width=4,height=4,units="in",bg="white",dpi=300)' % (cfg['out'] + '.' + pcol + '.qq_strat_mac.png')
elif cfg['ext'] == 'eps':
ggsave = 'ggsave(filename="%s",plot=gp,width=4,height=4,bg="white")' % (cfg['out'] + '.' + pcol + '.qq_strat_mac.eps')
else:
ggsave = 'ggsave(filename="%s",plot=gp,width=4,height=4,bg="white")' % (cfg['out'] + '.' + pcol + '.qq_strat_mac.pdf')
ro.r("""
gp<-ggplot(df, aes_string(x='expected',y='logp')) +
geom_point(aes_string(color='description'), size=2) +
scale_colour_manual(values=colorRampPalette(brewer.pal(9,"Blues"))(length(unique(df$description))+2)[3:(length(unique(df$description))+2)]) +
geom_abline(intercept=0, slope=1, alpha=0.5) +
scale_x_discrete(expression(Expected~~-log[10](italic(p)))) +
scale_y_discrete(expression(Observed~~-log[10](italic(p)))) +
coord_fixed() +
theme_bw(base_size = 12) +
theme(axis.title.x = element_text(vjust=-0.5,size=14), axis.title.y = element_text(vjust=1,angle=90,size=14), legend.title = element_blank(),
legend.key.height = unit(0.1,"in"), legend.text = element_text(size=6), legend.key = element_blank(), legend.justification = c(0,1),
legend.position = c(0,1), panel.background = element_blank(), panel.border = element_blank(), panel.grid.minor = element_blank(),
panel.grid.major = element_blank(), axis.line = element_line(colour="black"), axis.text = element_text(size=12))
%s
""" % (ggsave))
if np.max(results['logp']) > cfg['crop']:
print " generating cropped frequency stratified qq plot"
ro.r('df$logp[df$logp > ' + str(cfg['crop']) + ']<-' + str(cfg['crop']))
ro.r('df$shape<-0')
ro.r('df$shape[df$logp == ' + str(cfg['crop']) + ']<-1')
if cfg['ext'] == 'tiff':
ggsave = 'ggsave(filename="%s",plot=gp,width=4,height=4,units="in",bg="white",compression="lzw",dpi=300)' % (cfg['out'] + '.' + pcol + '.qq_strat_mac.cropped.tiff')
elif cfg['ext'] == 'png':
ggsave = 'ggsave(filename="%s",plot=gp,width=4,height=4,units="in",bg="white",dpi=300)' % (cfg['out'] + '.' + pcol + '.qq_strat_mac.cropped.png')
elif cfg['ext'] == 'eps':
ggsave = 'ggsave(filename="%s",plot=gp,width=4,height=4,bg="white")' % (cfg['out'] + '.' + pcol + '.qq_strat_mac.cropped.eps')
else:
ggsave = 'ggsave(filename="%s",plot=gp,width=4,height=4,bg="white")' % (cfg['out'] + '.' + pcol + '.qq_strat_mac.cropped.pdf')
ro.r("""
gp<-ggplot(df, aes_string(x='expected',y='logp')) +
geom_point(aes(shape=factor(shape), color=description), size=2) +
scale_colour_manual(values=colorRampPalette(brewer.pal(9,"Blues"))(length(unique(df$description))+2)[3:(length(unique(df$description))+2)]) +
geom_abline(intercept=0, slope=1, alpha=0.5) +
scale_x_discrete(expression(Expected~~-log[10](italic(p)))) +
scale_y_discrete(expression(Observed~~-log[10](italic(p)))) +
coord_fixed() +
theme_bw(base_size = 12) +
guides(shape=FALSE) +
theme(axis.title.x = element_text(vjust=-0.5,size=14), axis.title.y = element_text(vjust=1,angle=90,size=14), legend.title = element_blank(),
legend.key.height = unit(0.1,"in"), legend.text = element_text(size=6), legend.key = element_blank(), legend.justification = c(0,1),
legend.position = c(0,1), panel.background = element_blank(), panel.border = element_blank(), panel.grid.minor = element_blank(),
panel.grid.major = element_blank(), axis.line = element_line(colour="black"), axis.text = element_text(size=12))
%s
""" % (ggsave))
if cfg['mht']:
print " generating standard manhattan plot"
print " minimum p-value: " + str(np.min(results[pcol]))
print " maximum -1*log10(p-value): " + str(np.max(results['logp']))
if cfg['gc'] and l > 1:
print " adjusting p-values for genomic inflation for p-value column " + pcol
results[pcol]=2 * scipy.norm.cdf(-1 * np.abs(scipy.norm.ppf(0.5*results[pcol]) / math.sqrt(l)))
print " minimum post-gc adjustment p-value: " + str(np.min(results[pcol]))
print " maximum post-gc adjustment -1*log10(p-value): " + str(np.max(results['logp']))
else:
print " skipping genomic inflation correction"
print " calculating genomic positions"
results.sort_values(by=[cfg['chrcol'],cfg['bpcol']], inplace=True)
ticks = []
lastbase = 0
results['gpos'] = 0
nchr = len(list(np.unique(results[cfg['chrcol']].values)))
chrs = np.unique(results[cfg['chrcol']].values)
if cfg['color']:
colours = ["#08306B","#41AB5D","#000000","#F16913","#3F007D","#EF3B2C","#08519C","#238B45","#252525","#D94801","#54278F","#CB181D","#2171B5","#006D2C","#525252","#A63603","#6A51A3","#A50F15","#4292C6","#00441B","#737373","#7F2704","#807DBA","#67000D"]
else:
colours = ["#08589e","#4eb3d3","#08589e","#4eb3d3","#08589e","#4eb3d3","#08589e","#4eb3d3","#08589e","#4eb3d3","#08589e","#4eb3d3","#08589e","#4eb3d3","#08589e","#4eb3d3","#08589e","#4eb3d3","#08589e","#4eb3d3","#08589e","#4eb3d3","#08589e","#4eb3d3"]
if nchr == 1:
results['gpos'] = results[cfg['bpcol']]
results['colours'] = "#08589e"
if results['gpos'].max() - results['gpos'].min() <= 1000:
ticks = [x for x in range(results['gpos'].min(),results['gpos'].max()) if x % 100 == 0]
elif results['gpos'].max() - results['gpos'].min() <= 10000:
ticks = [x for x in range(results['gpos'].min(),results['gpos'].max()) if x % 1000 == 0]
elif results['gpos'].max() - results['gpos'].min() <= 100000:
ticks = [x for x in range(results['gpos'].min(),results['gpos'].max()) if x % 10000 == 0]
elif results['gpos'].max() - results['gpos'].min() <= 200000:
ticks = [x for x in range(results['gpos'].min(),results['gpos'].max()) if x % 20000 == 0]
elif results['gpos'].max() - results['gpos'].min() <= 300000:
ticks = [x for x in range(results['gpos'].min(),results['gpos'].max()) if x % 30000 == 0]
elif results['gpos'].max() - results['gpos'].min() <= 400000:
ticks = [x for x in range(results['gpos'].min(),results['gpos'].max()) if x % 40000 == 0]
elif results['gpos'].max() - results['gpos'].min() <= 500000:
ticks = [x for x in range(results['gpos'].min(),results['gpos'].max()) if x % 50000 == 0]
elif results['gpos'].max() - results['gpos'].min() <= 600000:
ticks = [x for x in range(results['gpos'].min(),results['gpos'].max()) if x % 60000 == 0]
elif results['gpos'].max() - results['gpos'].min() <= 700000:
ticks = [x for x in range(results['gpos'].min(),results['gpos'].max()) if x % 70000 == 0]
elif results['gpos'].max() - results['gpos'].min() <= 800000:
ticks = [x for x in range(results['gpos'].min(),results['gpos'].max()) if x % 80000 == 0]
elif results['gpos'].max() - results['gpos'].min() <= 900000:
ticks = [x for x in range(results['gpos'].min(),results['gpos'].max()) if x % 90000 == 0]
elif results['gpos'].max() - results['gpos'].min() <= 1000000:
ticks = [x for x in range(results['gpos'].min(),results['gpos'].max()) if x % 100000 == 0]
elif results['gpos'].max() - results['gpos'].min() <= 10000000:
ticks = [x for x in range(results['gpos'].min(),results['gpos'].max()) if x % 1000000 == 0]
elif results['gpos'].max() - results['gpos'].min() <= 100000000:
ticks = [x for x in range(results['gpos'].min(),results['gpos'].max()) if x % 10000000 == 0]
elif results['gpos'].max() - results['gpos'].min() > 100000000:
ticks = [x for x in range(results['gpos'].min(),results['gpos'].max()) if x % 25000000 == 0]
else:
results['colours'] = "#000000"
for i in range(len(chrs)):
print " processed chromosome " + str(int(chrs[i]))
if i == 0:
results.loc[results[cfg['chrcol']] == chrs[i],'gpos'] = results.loc[results[cfg['chrcol']] == chrs[i],cfg['bpcol']]
else:
lastbase = lastbase + results.loc[results[cfg['chrcol']] == chrs[i-1],cfg['bpcol']].iloc[-1]
results.loc[results[cfg['chrcol']] == chrs[i],'gpos'] = (results.loc[results[cfg['chrcol']] == chrs[i],cfg['bpcol']]) + lastbase
if results.loc[results[cfg['chrcol']] == chrs[i]].shape[0] > 1:
ticks.append(results.loc[results[cfg['chrcol']] == chrs[i],'gpos'].iloc[0] + (results.loc[results[cfg['chrcol']] == chrs[i],'gpos'].iloc[-1] - results.loc[results[cfg['chrcol']] == chrs[i],'gpos'].iloc[0])/2)
else:
ticks.append(results.loc[results[cfg['chrcol']] == chrs[i],'gpos'].iloc[0])
results.loc[results[cfg['chrcol']] == chrs[i],'colours'] = colours[int(chrs[i])]
results['logp'] = -1 * np.log10(results[pcol])
if results.shape[0] >= 1000000:
sig = 5.4e-8
else:
sig = 0.05 / results.shape[0]
print " significance level set to p-value = " + str(sig) + " (-1*log10(p-value) = " + str(-1 * np.log10(sig)) + ")"
print " " + str(len(results[pcol][results[pcol] <= sig])) + " genome wide significant variants"
chr = results[cfg['chrcol']][0]
maxy=int(max(np.ceil(-1 * np.log10(sig)),np.ceil(results['logp'].max())))
if maxy > 20:
y_breaks = range(0,maxy,5)
y_labels = range(0,maxy,5)
else:
y_breaks = range(0,maxy)
y_labels = range(0,maxy)
ro.globalenv['df'] = ro.DataFrame({'gpos': ro.FloatVector(results['gpos']), 'logp': ro.FloatVector(results['logp']), 'colours': ro.FactorVector(results['colours'])})
ro.globalenv['ticks'] = ro.FloatVector(ticks)
ro.globalenv['labels'] = ro.Vector(["{:,}".format(x/1000) for x in ticks])
ro.globalenv['colours'] = ro.StrVector(colours)
ro.globalenv['chrs'] = ro.FloatVector(chrs)
print " generating manhattan plot"
if cfg['ext'] == 'tiff':
ggsave = 'ggsave(filename="%s",plot=gp,width=16,height=4,units="in",bg="white",compression="lzw",dpi=300)' % (cfg['out'] + '.' + pcol + '.mht.tiff')
elif cfg['ext'] == 'png':
ggsave = 'ggsave(filename="%s",plot=gp,width=16,height=4,units="in",bg="white",dpi=300)' % (cfg['out'] + '.' + pcol + '.mht.png')
elif cfg['ext'] == 'eps':
ggsave = 'ggsave(filename="%s",plot=gp,width=16,height=4,bg="white")' % (cfg['out'] + '.' + pcol + '.mht.eps')
else:
ggsave = 'ggsave(filename="%s",plot=gp,width=16,height=4,bg="white")' % (cfg['out'] + '.' + pcol + '.mht.pdf')
if nchr == 1:
ro.r("""
gp<-ggplot(df, aes_string(x='gpos',y='logp')) +
geom_hline(yintercept = -1 * log10(%g),colour="#B8860B", linetype=5, size = 0.25) +
geom_point(size=1.5) +
scale_x_continuous(expression(Chromosome~~%d~~(kb))'),breaks=ticks,labels=labels) + \
scale_y_continuous(expression(-log[10](italic(p))),breaks=seq(0,%d,1),limits=c(0,%d)) + \
theme_bw(base_size = 8) + \
theme(axis.title.x = element_text(vjust=-0.5,size=14), axis.title.y = element_text(vjust=1,angle=90,size=14),
panel.background = element_blank(), panel.border = element_blank(), panel.grid.minor = element_blank(),
panel.grid.major = element_blank(), axis.line = element_line(colour="black"), axis.title = element_text(size=10),
axis.text = element_text(size=12), legend.position = 'none')
%s
""" % (sig, chr, maxy, maxy, ggsave))
else:
ro.r("""
gp = ggplot(df, aes_string(x='gpos',y='logp',colour='colours')) +
geom_hline(yintercept = -1 * log10(%g),colour="#B8860B", linetype=5, size = 0.25) +
geom_point(size=1.5) +
scale_colour_manual(values=colours) +
scale_x_continuous(expression(Chromosome),breaks=ticks,labels=chrs) +
scale_y_continuous(expression(-log[10](italic(p))),breaks=seq(0,%d,1),limits=c(0,%d)) +
theme_bw(base_size = 8) +
theme(axis.title.x = element_text(vjust=-0.5,size=14), axis.title.y = element_text(vjust=1,angle=90,size=14),
panel.background = element_blank(), panel.border = element_blank(), panel.grid.minor = element_blank(),
panel.grid.major = element_blank(), axis.line = element_line(colour="black"), axis.title = element_text(size=10),
axis.text = element_text(size=12), legend.position = 'none')
%s
""" % (sig, maxy, maxy, ggsave))
if maxy > cfg['crop']:
maxy = cfg['crop']
ro.r('df$logp[df$logp > ' + str(cfg['crop']) + ']<-' + str(cfg['crop']))
ro.r('df$shape<-0')
ro.r('df$shape[df$logp == ' + str(cfg['crop']) + ']<-1')
print " generating cropped manhattan plot"
if cfg['ext'] == 'tiff':
ggsave = 'ggsave(filename="%s",plot=gp,width=16,height=4,units="in",bg="white",compression="lzw",dpi=300)' % (cfg['out'] + '.' + pcol + '.mht.cropped.tiff')
elif cfg['ext'] == 'png':
ggsave = 'ggsave(filename="%s",plot=gp,width=16,height=4,units="in",bg="white",dpi=300)' % (cfg['out'] + '.' + pcol + '.mht.cropped.png')
elif cfg['ext'] == 'eps':
ggsave = 'ggsave(filename="%s",plot=gp,width=16,height=4,bg="white")' % (cfg['out'] + '.' + pcol + '.mht.cropped.eps')
else:
ggsave = 'ggsave(filename="%s",plot=gp,width=16,height=4,bg="white")' % (cfg['out'] + '.' + pcol + '.mht.cropped.pdf')
if nchr == 1:
ro.r("""
gp<-ggplot(df, aes_string(x='gpos',y='logp')) +
geom_hline(yintercept = -1 * log10(%g),colour="#B8860B", linetype=5, size = 0.25) +
geom_point(aes(shape=factor(shape)),size=1.5) +
scale_x_continuous(expression(Chromosome~~%d~~(kb))'),breaks=ticks,labels=labels) +
scale_y_continuous(expression(-log[10](italic(p))),breaks=seq(0,%d,1),limits=c(0,%d)) +
theme_bw(base_size = 8) +
theme(axis.title.x = element_text(vjust=-0.5,size=14), axis.title.y = element_text(vjust=1,angle=90,size=14),
panel.background = element_blank(), panel.border = element_blank(), panel.grid.minor = element_blank(),
panel.grid.major = element_blank(), axis.line = element_line(colour="black"), axis.title = element_text(size=10),
axis.text = element_text(size=12), legend.position = 'none')
%s
""" % (sig, chr, maxy, maxy, ggsave))
else:
ro.r("""
gp = ggplot(df, aes_string(x='gpos',y='logp',colour='colours')) +
geom_hline(yintercept = -1 * log10(%g),colour="#B8860B", linetype=5, size = 0.25) +
geom_point(aes(shape=factor(shape)),size=1.5) +
scale_colour_manual(values=colours) +
scale_x_continuous(expression(Chromosome),breaks=ticks,labels=chrs) +
scale_y_continuous(expression(-log[10](italic(p))),breaks=seq(0,%d,1),limits=c(0,%d)) +
theme_bw(base_size = 8) +
theme(axis.title.x = element_text(vjust=-0.5,size=14), axis.title.y = element_text(vjust=1,angle=90,size=14),
panel.background = element_blank(), panel.border = element_blank(), panel.grid.minor = element_blank(),
panel.grid.major = element_blank(), axis.line = element_line(colour="black"), axis.title = element_text(size=8),
axis.text = element_text(size=12), legend.position = 'none')
%s
""" % (sig, maxy, maxy, ggsave))
print "process complete"
return 0
def process_vcf_into_selscan_tped(cls, vcf_file, gen_map_file, outfile_location,
outfile_prefix, chromosome_num, samples_to_include=None, start_pos_bp=None, end_pos_bp=None, ploidy=2,
consider_multi_allelic=True, rescale_genetic_distance=False, include_variants_with_low_qual_ancestral=False, coding_function=None,
multi_alleli_merge_function="AND"):
"""
Process a bgzipped-VCF (such as those included in the Phase 3 1000 Genomes release) into a gzip-compressed
tped file of the sort expected by selscan.
"""
assert ploidy > 0
processor = VCFReader(vcf_file)
end_pos = processor.clens[str(chromosome_num)] if end_pos_bp is None else end_pos_bp
records = processor.records( str(chromosome_num), start_pos_bp, end_pos, pysam.asVCF())
outTpedFile = outfile_location + "/" + outfile_prefix + ".tped.gz"
outTpedMetaFile = outfile_location + "/" + outfile_prefix + ".tped.allele_metadata.gz"
if samples_to_include is not None and len(samples_to_include) > 0:
indices_of_matching_samples = sorted([processor.sample_names.index(x) for x in samples_to_include])
else:
indices_of_matching_samples = range(0,len(processor.sample_names))
indices_of_matching_genotypes = [(x*2, (x*2)+1) for x in indices_of_matching_samples]
indices_of_matching_genotypes = list(np.ravel(np.array(indices_of_matching_genotypes)))
rm = RecomMap(gen_map_file)
for filePath in [outTpedFile, outTpedMetaFile]:
assert not os.path.exists(filePath), "File {} already exists. Consider removing this file or specifying a different output prefix. Processing aborted.".format(filePath)
mergeOperatorString = ""
if multi_alleli_merge_function == "OR":
mergeOperatorString = "|"
if multi_alleli_merge_function == "AND":
mergeOperatorString = "&"
if multi_alleli_merge_function == "XOR":
mergeOperatorString = "^"
startTime = datetime.now()
sec_remaining_avg = 0
current_pos_bp = 1
with util.file.open_or_gzopen(outTpedFile, 'w') as of1, util.file.open_or_gzopen(outTpedMetaFile, 'w') as of2:
# WRITE header for metadata file here with selected subset of sample_names
headerString = "CHROM VARIANT_ID POS_BP MAP_POS_CM REF_ALLELE ALT_ALLELE ANCESTRAL_CALL ALLELE_FREQ_IN_POP\n".replace(" ","\t")
of2.write(headerString)
of1linesToWrite = []
of2linesToWrite = []
recordLengths = set() #
recordCount = 0
mostRecentRecordPosSeen = -1
positionHasBeenSeenBefore = False
previouslyCodedGenotypes = GenoRecord([])
lineToWrite1 = None
lineToWrite2 = None
previousAncestral = None
ancestralDiffersFromPrevious = True
for record in records:
# in some cases, there may be records with duplicate positions in the VCF file
# to account for that we collapse rows that pass our filter and then write out the rows when we
# encounter a record with a new position
if record.pos != mostRecentRecordPosSeen:
if positionHasBeenSeenBefore and not consider_multi_allelic:
lineToWrite1 = None
lineToWrite2 = None
if lineToWrite1 is not None and lineToWrite2 is not None:
if len(previouslyCodedGenotypes) == ploidy*len(indices_of_matching_samples):
# write the output line
of1.write(lineToWrite1)
of2.write(lineToWrite2)
else:
genotypesCount = len(previouslyCodedGenotypes)
countSpecificTpedName = outTpedFile.replace(outfile_prefix, outfile_prefix + "_" + str(genotypesCount))
countSpecificMetafileName = outTpedMetaFile.replace(outfile_prefix, outfile_prefix + "_" + str(genotypesCount))
with util.file.open_or_gzopen(countSpecificTpedName, 'a') as of1l, util.file.open_or_gzopen(countSpecificMetafileName, 'a') as of2l:
of1l.write(lineToWrite1)
of2l.write(lineToWrite2)
lineToWrite1 = None
lineToWrite2 = None
mostRecentRecordPosSeen = record.pos
positionHasBeenSeenBefore = False
else:
positionHasBeenSeenBefore = True
# if the variant is a SNP
# OLD style looking at INFO VT value:
# processor.variant_is_type(record.info, "SNP"):
VALID_BASES = ["A","C","G","T","N","a","c","g","t","n"]
if (len(record.ref) == 1 and len(record.alt) == 1) or ( all(variant in VALID_BASES for variant in record.ref.split(",")) and
all(variant in VALID_BASES for variant in record.alt.split(",")) ):
alternateAlleles = [record.alt]
if record.alt not in ['A','T','C','G']:
#print record.alt
if consider_multi_allelic:
pass
alternateAlleles = record.alt.split(",")
else:
# continue on to next variant record
continue
ancestral_allele = processor.parse_ancestral(record.info)
chromStr = "chr{}".format(record.contig)
# if the AA is populated, and the call meets the specified criteria
if (ancestral_allele in ['A','T','C','G']) or (include_variants_with_low_qual_ancestral and ancestral_allele in ['a','t','c','g']):
if previousAncestral != ancestral_allele:
previousAncestral = ancestral_allele
ancestralDiffersFromPrevious = True
else:
ancestralDiffersFromPrevious = False
recordString = record.__str__()
match = cls.genoRegex.match(recordString)
if match:
rawGenos = match.group("genos")
genos = rawGenos[::2]
recordPosition = record.pos+1
if chromosome_num.upper() != "X":
try:
genotypes_for_selected_samples = operator.itemgetter(*indices_of_matching_genotypes)(genos)
except Exception: # if this is a record of mixed ploidy, that is to say the X chromosome
raise
else:
matching_genotypes = np.array(record[:len(record)])[indices_of_matching_samples]
genotypes_for_selected_samples_split = [x.split("|") for x in matching_genotypes]
genotypes_for_selected_samples = [y for x in genotypes_for_selected_samples_split for y in x]
recordLengths.add(len(genotypes_for_selected_samples))
map_pos_cm = rm.physToMap(chromStr, record.pos, rescale=rescale_genetic_distance)
numberOfHaplotypes = float(len(genotypes_for_selected_samples))
codingFunc = np.vectorize(coding_function)
coded_genotypes_for_selected_samples = GenoRecord(["0"] * len(genotypes_for_selected_samples))
if consider_multi_allelic:
#coded_genotypes_for_selected_samples = GenoRecord(["1"] * len(genotypes_for_selected_samples))
for idx, altAllele in enumerate(alternateAlleles):
value_of_current_allele = str(idx+1)
coded_genotypes_for_selected_samples_for_allele = GenoRecord(codingFunc(genotypes_for_selected_samples, record.ref, altAllele, ancestral_allele, value_of_current_allele))
#coded_genotypes_for_selected_samples |= coded_genotypes_for_selected_samples_for_allele
if idx==0:
coded_genotypes_for_selected_samples = coded_genotypes_for_selected_samples_for_allele
else:
coded_genotypes_for_selected_samples = coded_genotypes_for_selected_samples.f[mergeOperatorString](coded_genotypes_for_selected_samples_for_allele)
#coded_genotypes_for_selected_samples = np.array(list(str(bin(int("".join(coded_genotypes_for_selected_samples),2) | int("".join(coded_genotypes_for_selected_samples_for_allele),2)))[2:].zfill(len(coded_genotypes_for_selected_samples))))
else:
coded_genotypes_for_selected_samples = GenoRecord(codingFunc(genotypes_for_selected_samples, record.ref, record.alt, ancestral_allele, "1"))
# if this is the first record in the file, create an array filled with zeros for the previously coded alleles
if recordCount == 0:
previouslyCodedGenotypes = GenoRecord(["1"] * len(genotypes_for_selected_samples))
# bitwise OR coded genotypes for duplicate records, merge variants
# ...except selscan logic is inverted, so bitwise AND
# TODO: invert?
# record @ pos1 = 001001
# record @ pos1 = 100001
# -------
# coded result = 101001
#log.debug(genotypes_for_selected_samples)
#log.debug(coded_genotypes_for_selected_samples)
if positionHasBeenSeenBefore:
#coded_genotypes_for_selected_samples |= previouslyCodedGenotypes
coded_genotypes_for_selected_samples = coded_genotypes_for_selected_samples.f[mergeOperatorString](previouslyCodedGenotypes)
#coded_genotypes_for_selected_samples = np.array(list(str(bin(int("".join(coded_genotypes_for_selected_samples),2) & int("".join(previouslyCodedGenotypes),2)))[2:].zfill(len(coded_genotypes_for_selected_samples))))
previouslyCodedGenotypes = coded_genotypes_for_selected_samples
allele_freq_for_pop = float(list(coded_genotypes_for_selected_samples).count("1")) / numberOfHaplotypes
outStrDict = cls._build_variant_output_strings(record.contig, str(1), recordPosition,
map_pos_cm, coded_genotypes_for_selected_samples, record.ref, record.alt,
ancestral_allele, allele_freq_for_pop)
lineToWrite1 = outStrDict["tpedString"]
lineToWrite2 = outStrDict["metadataString"].replace(" ","\t")
recordCount += 1
current_pos_bp = int(recordPosition)
if recordCount % 1000 == 0:
number_of_seconds_elapsed = (datetime.now() - startTime).total_seconds()
bp_per_sec = float(current_pos_bp) / float(number_of_seconds_elapsed)
bp_remaining = end_pos - current_pos_bp
sec_remaining = bp_remaining / bp_per_sec
sec_remaining_avg = cls._moving_avg(sec_remaining, sec_remaining_avg, 10)
time_left = timedelta(seconds=sec_remaining_avg)
if sec_remaining > 10:
human_time_remaining = relative_time(datetime.utcnow()+time_left)
print("")
print("Completed: {:.2%}".format(float(current_pos_bp)/float(end_pos)))
print("Estimated time of completion: {}".format(human_time_remaining))
#log.info("Genotype counts found: %s", str(list(recordLengths)))
if positionHasBeenSeenBefore and not consider_multi_allelic:
lineToWrite1 = None
lineToWrite2 = None
if lineToWrite1 is not None and lineToWrite2 is not None:
if len(previouslyCodedGenotypes) == ploidy*len(indices_of_matching_samples):
# write the output lines
of1.write(lineToWrite1)
of2.write(lineToWrite2)
else:
genotypesCount = len(previouslyCodedGenotypes)
countSpecificTpedName = outTpedFile.replace(outfile_prefix, outfile_prefix + "_" + str(genotypesCount))
countSpecificMetafileName = outTpedMetaFile.replace(outfile_prefix, outfile_prefix + "_" + str(genotypesCount))
with util.file.open_or_gzopen(countSpecificTpedName, 'a') as of1l, util.file.open_or_gzopen(countSpecificMetafileName, 'a') as of2l:
of1l.write(lineToWrite1)
of2l.write(lineToWrite2)
log.info("Genotype counts found: %s", str(list(recordLengths)))
def __init__(self, vcf_files):
self._readers = []
for file_name in vcf_files:
self._readers.append(
pysam.Tabixfile(file_name, parser=pysam.asVCF()))
annos[anno] = BigWigFile(open(anno_files[anno]))
except IOError:
sys.exit("Gemini cannot open this annotation file: %s. \n"
"Have you installed the annotation files? If so, "
"have they been moved or deleted? Exiting...\n\n"
"For more details:\n\t"
"http://gemini.readthedocs.org/en/latest/content/"
"#installation.html\#installing-annotation-files\n"
% anno_files[anno])
# ## Standard access to Tabix indexed files
PARSERS = {"bed": pysam.asBed(),
"vcf": pysam.asVCF(),
"tuple": pysam.asTuple(),
None: None}
def _get_hits(coords, annotation, parser_type, _parsers=PARSERS):
"""Retrieve BED information, recovering if BED annotation file does have a chromosome.
"""
try:
parser = _parsers[parser_type]
except KeyError:
raise ValueError("Unexpected parser type: %s" % parser)
chrom, start, end = coords
try:
hit_iter = annotation.fetch(str(chrom), start, end, parser=parser)
# catch invalid region errors raised by ctabix
except ValueError:
def handle(
self,
file: str,
organism: str,
doi: str = None,
cpu: int = 1,
verbosity: int = 1,
**options
):
"""Execute the main function."""
# retrieve only the file name
filename = os.path.basename(file)
if verbosity > 0:
self.stdout.write("Processing file: {}".format(filename))
try:
FileValidator().validate(file)
except ImportingError as e:
raise CommandError(e)
try:
index_file = "{}.tbi".format(file)
FileValidator().validate(index_file)
except ImportingError:
try:
index_file = "{}.csi".format(file)
FileValidator().validate(index_file)
except ImportingError:
raise CommandError("No index found (.tbi/.csi)")
try:
feature_file = FeatureLoader(
filename=filename, source="VCF_SOURCE", doi=doi
)
except ImportingError as e:
raise CommandError(e)
pool = ThreadPoolExecutor(max_workers=cpu)
tasks = list()
chunk_size = cpu * 2
# Load the GFF3 file
with open(file) as tbx_file:
tbx = pysam.TabixFile(filename=tbx_file.name, index=index_file)
for row in tqdm(tbx.fetch(parser=pysam.asVCF()), total=get_num_lines(file)):
tasks.append(
pool.submit(feature_file.store_tabix_VCF_feature, row, organism)
)
if len(tasks) >= chunk_size:
for task in as_completed(tasks):
try:
task.result()
except ImportingError as e:
raise CommandError(e)
tasks.clear()
else:
for task in as_completed(tasks):
try:
task.result()
except ImportingError as e:
raise CommandError(e)
tasks.clear()
pool.shutdown()
if verbosity > 0:
self.stdout.write(self.style.SUCCESS("Done with {}".format(filename)))
def Variants(self, chromosome, start, end, freqAsPrior=False):
"""
Generator funtion. Yields variants in order of
genomic co-ordinate.
"""
varList = []
for vcfFile in self.vcfFiles:
try:
vcfLines = vcfFile.fetch(chromosome,
start,
end,
parser=pysam.asVCF())
except Exception, e:
logger.warning(
"Could not retrieve variants from source file in region %s:%s-%s. Error was %s"
% (chromosome, start, end, e.message))
continue
for line in vcfLines:
pos = line.pos
ref = line.ref
alt = line.alt
info = line.info
infoEntries = info.split(";")
freq = None
if freqAsPrior:
for entry in infoEntries:
cols = entry.split("=")
if cols[0] == "AF":
if len(cols) == 2:
freq = float(cols[1])
else:
# Skip multi-allelic sites for now
continue
lenRef = len(ref)
lenAlt = len(alt)
# SNP
if lenRef == 1 and lenAlt == 1:
var = Variant(chromosome, pos, ref, alt, 0, 0, 0, 0, 0)
if freqAsPrior:
var.freqPrior = freq
varList.append(var)
# Anything else
else:
# VCF4 is -1 indexed for indels, so trim off first base
ref = ref[1:]
alt = alt[1:]
removed = ref
added = alt
# Trim the matching bits off and shift position. This will decompose
# multi-variant sites into individual alleles at different positions.
while len(ref) > 0 and len(alt) > 0 and ref[0] == alt[0]:
ref = ref[1:]
alt = alt[1:]
removed = ref
added = alt
pos += 1
var = Variant(chromosome, pos, removed, added, 0, 0, 0, 0,
0)
if freqAsPrior:
var.freqPrior = freq
varList.append(var)
def process_piece(filename_vcf, chrom, chrom_length, sample_index, chain_info,
diploid, passed, quality, vcf_keep, vcf_discard_file):
ret = {'chrom': chrom, 'stats': {}, 'chain_info': chain_info}
stats = OrderedDict()
stats['ACCEPTED'] = 0
if vcf_keep:
vcf_discard = open(vcf_discard_file, "w")
line_no = 0
try:
LOG.info("Processing Chromosome {0}...".format(chrom))
tb = pysam.TabixFile(filename_vcf)
for vcf_rec in tb.fetch(chrom, parser=pysam.asVCF()):
line_no += 1
try:
gt = parse_gt_new(vcf_rec, sample_index)
except:
LOG.info("Unable to parse record, improper VCF file?")
continue
LOG.debug('\n')
LOG.debug(vcf_rec)
LOG.debug(gt)
LOG.debug(vcf_rec[sample_index])
if passed and 'PASS' not in vcf_rec.FILTER:
LOG.debug("TOSSED: FILTERED ON PASS")
stats = update_stats(stats, 'FILTERED ON PASS')
if vcf_keep:
vcf_discard.write(vcf_rec)
vcf_discard.write("\n")
continue
elif quality and gt.fi == '0':
# FI : Whether a sample was a Pass(1) or fail (0) based on FILTER values
LOG.debug("TOSSED: FILTERED ON QUALITY")
stats = update_stats(stats, 'FILTERED ON QUALITY')
if vcf_keep:
vcf_discard.write(vcf_rec)
vcf_discard.write("\n")
continue
elif gt.left is None and gt.right is None:
LOG.debug("TOSSED: NOT RELEVANT")
stats = update_stats(stats, 'NOT RELEVANT')
if vcf_keep:
vcf_discard.write(vcf_rec)
vcf_discard.write("\n")
continue
elif not diploid and gt.left != gt.right:
# haploid or hexaploid
# gt must be equal
LOG.debug("TOSSED: HETEROZYGOUS")
stats = update_stats(stats, 'HETEROZYGOUS')
if vcf_keep:
vcf_discard.write(vcf_rec)
vcf_discard.write("\n")
continue
# START L AND R, ONLY R IF DIPLOID
for ci, lr in chain_info.iteritems():
if ci == 'left':
alt_seq = str(gt.left)
else:
alt_seq = str(gt.right)
if gt.ref == alt_seq:
LOG.debug("TOSSED, SAME AS REF")
lr.stats = update_stats(lr.stats, 'SAME AS REF')
if vcf_keep:
vcf_discard.write(vcf_rec)
vcf_discard.write("\n")
continue
orig_alt_seq = alt_seq
LOG.debug("SAMPLE: {0}".format(vcf_rec[sample_index]))
LOG.debug(
"REF='{0}', ALT_L='{1}', ALT_R='{2}'. POS={3}".format(
gt.ref, gt.left, gt.right, vcf_rec.pos))
position = vcf_rec.pos + 1
ref_seq = str(gt.ref)
len_ref = len(ref_seq)
len_alt = len(alt_seq)
base_changes = len_ref - len_alt
base_pos_diff = 0
if position < lr.prev_next_ref_pos:
LOG.debug(
"TOSSED: CONFLICTING VCF ENTRIES: {0}".format(vcf_rec))
lr.stats = update_stats(lr.stats,
'CONFLICTING VCF ENTRIES')
if vcf_keep:
vcf_discard.write(vcf_rec)
vcf_discard.write("\n")
continue
# find the position where the first base change is
for n in xrange(min(len_ref, len_alt)):
if ref_seq[n] != alt_seq[n]:
base_pos_diff = n
break
# if it is 0, take the minimum length
if base_pos_diff == 0:
base_pos_diff = min(len_ref, len_alt)
# add the base position difference
position += base_pos_diff
# recalculate the strings
shared_bases = ref_seq[:base_pos_diff]
ref_seq = ref_seq[base_pos_diff:]
alt_seq = alt_seq[base_pos_diff:]
dt = len(ref_seq)
dq = len(alt_seq)
next_ref_pos = position + len(ref_seq)
fragment_size = position - lr.prev_next_ref_pos
LOG.debug(' gt.ref: {0}'.format(gt.ref))
LOG.debug(' ref_seq: {0}'.format(ref_seq))
LOG.debug(' dt: {0}'.format(dt))
LOG.debug(' gt.alt: {0}'.format(orig_alt_seq))
LOG.debug(' alt_seq: {0}'.format(alt_seq))
LOG.debug(' dq: {0}'.format(dq))
LOG.debug(' position: {0}'.format(position))
LOG.debug('prev_next_ref_pos: {0}'.format(
lr.prev_next_ref_pos))
LOG.debug(' next_ref_pos: {0}'.format(next_ref_pos))
LOG.debug(' fragment_size: {0}'.format(fragment_size))
LOG.debug(' base_changes: {0}'.format(base_changes))
LOG.debug(' base_pos_diff: {0}'.format(base_pos_diff))
LOG.debug(' shared_bases: {0}'.format(shared_bases))
# fix any 0 length
if fragment_size < 0:
LOG.debug(
"TOSSED: CONFLICTING VCF ENTRIES: {0}".format(vcf_rec))
lr.stats = update_stats(lr.stats,
'CONFLICTING VCF ENTRIES')
if vcf_keep:
vcf_discard.write(vcf_rec)
vcf_discard.write("\n")
continue
if fragment_size != 0:
ref_str = ref_seq if ref_seq else '.'
alt_str = alt_seq if alt_seq else '.'
lr.chain_entries.append([
fragment_size,
len(ref_seq),
len(alt_seq), shared_bases, ref_str, alt_str,
vcf_rec.pos + 1
])
else:
#
# THIS SHOULD NOT HAPPEN
#
raise G2GChainFileError(
'Unable to create chain file due to conflicting VCF entries'
)
lr.stats = update_stats(lr.stats, 'ACCEPTED')
LOG.debug(lr.chain_entries[-1])
last_position = position
lr.prev_next_ref_pos = next_ref_pos
lr.sums[0] += fragment_size
lr.sums[1] += dt
lr.sums[2] += dq
prev_line = vcf_rec
#lr.prev_chrom = vcf_rec.contig
chain_info[ci] = lr
for ci, lr in chain_info.iteritems():
#LOG.debug("CHROMOSOME[{0}] LENGTH = {1}".format(lr.prev_chrom, chrom_length))
lr.chromosome = chrom
lr.chromosome_length = chrom_length
lr.last_fragment_size = chrom_length - lr.sums[0] - lr.sums[1]
lr.end_length = lr.sums[0] + lr.last_fragment_size + lr.sums[2]
lr.number_vcf_lines = line_no
chain_info[ci] = lr
if vcf_keep:
vcf_discard.close()
except KeyboardInterrupt:
raise KeyboardInterruptError()
except Exception, e:
pass
subprocess.check_output("samtools faidx {}/ref.fa".format(outdir), shell=True)
logger.info("Lifter over the reference to %s/ref.fa" % outdir)
# This will liftover the VCFs to the new reference
for invcf in invcfs:
if not os.path.isfile(invcf):
logger.error("%s not found" % invcf)
continue
# get the base name and use it in the output
outvcf = os.path.join(outdir, os.path.splitext(os.path.basename(invcf))[0])
vcf_template_reader = vcf.Reader(open(invcf, "r"))
vcf_template_reader.metadata["reference"] = os.path.join(outdir, "ref.fa")
vcf_template_reader.contigs = OrderedDict([(contig_name, vcf.parser._Contig(contig_name, contig_length)) for (contig_name, contig_length) in contigs])
vcf_writer = vcf.Writer(open(outvcf, "w"), vcf_template_reader)
tabix_vcf = pysam.TabixFile(invcf, parser=pysam.asVCF())
info_warned = False
for region_index, region in enumerate(regions_bedtool, start=1):
records = None
try: records = tabix_vcf.fetch(reference=str(region.chrom), start=region.start, end=region.end)
except ValueError: logger.error("Failed to retrieve %s from %s" % (str(region).strip(), invcf))
if records is None: continue
for record in records:
if record.pos <= region.start + args.flank or record.pos + len(record.ref) + args.flank - 1 >= region.end: continue
record.contig = str(region_index) if args.short_contig_names else ("%s_%d_%d" % (str(region.chrom), region.start, region.end))
# record.pos seems to be zero-based, at least in the infinite wisdom of my version of pysam
record.pos = record.pos - region.start
fmt = "GT" if len(record) else None
sample_indexes = [0] if len(record) else []
info = None
try: info = vcf_template_reader._parse_info(record.info)
