def find_nocov_variants(covlist,chrom='',caller='',min_cov=5):
variants = []
assert min(covlist[1:]) >= 0
nocov = [i for i,v in enumerate(covlist) if v < min_cov]
nocov.remove(0) # take off the -1 at index 0
if len(covlist)-1 == len(nocov): return None # entire sequence has no coverage
nocov_intervals = list(intervals(nocov))
for iv in nocov_intervals:
data = {'chrom':chrom,'caller':caller,'pos':iv[0], 'type': 'no_cov'}
data['length'] = iv[1] - iv[0] + 1
data['mean_cov'] = scipy.mean(covlist[iv[0]:(iv[1]+1)])
variants.append(Variant.from_dict(data))
return variants
def find_variants(covlist,
seq,
chrom,
min_cov=5,
min_score=30,
exclude_edges=False,
exclude_overlaps=False):
''' identify coverage variants in covlist
Returns dict with keys 'mean_cov','pct_cov', and 'variants', where dict['variants']
is a list of Variant objects
'''
assert min(covlist[1:]) >= 0
assert len(covlist) - 1 == len(
seq
), "Number of coverage values (%d) is not equal to sequence length (%d)" % (
len(covlist) - 1, len(seq))
retval = {}
nocov = [i for i, v in enumerate(covlist) if v < min_cov]
nocov.remove(0)
retval['mean_cov'] = scipy.mean(covlist[1:])
retval['pct_cov'] = 1 - (float(len(nocov)) / (len(covlist) - 1))
if len(nocov) == len(seq):
return retval
nocov_intervals = list(intervals(nocov))
#covscores,localmeans = local_coverage_score(covlist)
covscores, localmeans = adjusted_coverage_score(covlist)
covdip = [i for i, v in enumerate(covscores) if v >= min_score]
covdip_intervals = list(intervals(covdip))
# refine intervals
if exclude_edges:
# ignore intervals that overlap the beginning and end of reference
covdip_intervals = [
iv for iv in covdip_intervals
if not iv[0] == 1 and not iv[1] == (len(covlist) - 1)
]
if exclude_overlaps:
# ignore covdip intervals that overlap with nocov intervals
covdip_intervals = remove_overlap(covdip_intervals, nocov_intervals)
# covdip = list(itertools.chain(*[range(v1,v2+1) for v1,v2 in covdip_intervals])
# positions with no coverage are not considered to be coverage dips
covdip = [p for p in covdip if p not in nocov]
variants = []
for iv in nocov_intervals:
data = {'chrom': chrom, 'pos': iv[0], 'type': 'no_cov'}
data['length'] = iv[1] - iv[0] + 1
data['mean_cov'] = scipy.mean(covlist[iv[0]:(iv[1] + 1)])
variants.append(Variant.from_dict(data))
for iv in covdip_intervals:
data = {'chrom': chrom, 'pos': iv[0], 'type': 'cov_dip'}
data['length'] = iv[1] - iv[0] + 1
data['mean_cov'] = scipy.mean(covlist[iv[0]:(iv[1] + 1)])
intscores = covscores[iv[0]:(iv[1] + 1)]
intmeans = localmeans[iv[0]:(iv[1] + 1)]
data['quality'] = max(intscores)
data['info'] = {
'CovScores':
'%s' % ','.join(['%d' % int(round(v)) for v in intscores]),
'LocalMeans':
'%s' % ','.join(['%d' % int(round(v)) for v in intmeans]),
}
data['ref'] = str(seq[iv[0]:(iv[1] + 1)].seq).upper()
# data['alt'] = data['ref'].lower()
variants.append(Variant.from_dict(data))
if variants:
retval['variants'] = variants
return retval
def find_variants(covlist, seq, chrom, min_cov=5, min_score=30, exclude_edges=False, exclude_overlaps=False):
''' identify coverage variants in covlist
Returns dict with keys 'mean_cov','pct_cov', and 'variants', where dict['variants']
is a list of Variant objects
'''
assert min(covlist[1:]) >= 0
assert len(covlist) - 1 == len(seq), "Number of coverage values (%d) is not equal to sequence length (%d)" % (len(covlist)-1,len(seq))
retval = {}
nocov = [i for i,v in enumerate(covlist) if v < min_cov]
nocov.remove(0)
retval['mean_cov'] = scipy.mean(covlist[1:])
retval['pct_cov'] = 1 - (float(len(nocov)) / (len(covlist) - 1))
if len(nocov) == len(seq):
return retval
nocov_intervals = list(intervals(nocov))
#covscores,localmeans = local_coverage_score(covlist)
covscores,localmeans = adjusted_coverage_score(covlist)
covdip = [i for i,v in enumerate(covscores) if v >= min_score]
covdip_intervals = list(intervals(covdip))
# refine intervals
if exclude_edges:
# ignore intervals that overlap the beginning and end of reference
covdip_intervals = [iv for iv in covdip_intervals if not iv[0]==1 and not iv[1]==(len(covlist)-1)]
if exclude_overlaps:
# ignore covdip intervals that overlap with nocov intervals
covdip_intervals = remove_overlap(covdip_intervals,nocov_intervals)
# covdip = list(itertools.chain(*[range(v1,v2+1) for v1,v2 in covdip_intervals])
# positions with no coverage are not considered to be coverage dips
covdip = [p for p in covdip if p not in nocov]
variants = []
for iv in nocov_intervals:
data = {'chrom':chrom, 'pos':iv[0], 'type': 'no_cov'}
data['length'] = iv[1] - iv[0] + 1
data['mean_cov'] = scipy.mean(covlist[iv[0]:(iv[1]+1)])
variants.append(Variant.from_dict(data))
for iv in covdip_intervals:
data = {'chrom':chrom, 'pos':iv[0], 'type': 'cov_dip'}
data['length'] = iv[1] - iv[0] + 1
data['mean_cov'] = scipy.mean(covlist[iv[0]:(iv[1]+1)])
intscores = covscores[iv[0]:(iv[1]+1)]
intmeans = localmeans[iv[0]:(iv[1]+1)]
data['quality'] = max(intscores)
data['info'] = {'CovScores':'%s' % ','.join(['%d' % int(round(v)) for v in intscores]),
'LocalMeans':'%s' % ','.join(['%d' % int(round(v)) for v in intmeans]),
}
data['ref'] = str(seq[iv[0]:(iv[1]+1)].seq).upper()
# data['alt'] = data['ref'].lower()
variants.append(Variant.from_dict(data))
if variants:
retval['variants'] = variants
return retval