def euclidean_distance(sites_a, sites_b):
"""Euclidean distance between two sets of sites"""
ma = bioutils.build_motif(sites_a)
mb = bioutils.build_motif(sites_b)
def ed(cola, colb):
return math.sqrt(sum((cola[l] - colb[l])**2 for l in "ACGT"))
return sum(ed(cola, colb) for (cola,colb) in zip(ma.pwm(), mb.pwm()))
def euclidean_distance(sites_a, sites_b):
"""Euclidean distance between two sets of sites"""
ma = bioutils.build_motif(sites_a)
mb = bioutils.build_motif(sites_b)
def ed(cola, colb):
return math.sqrt(sum((cola[l] - colb[l])**2 for l in "ACGT"))
return sum(ed(cola, colb) for (cola, colb) in zip(ma.pwm(), mb.pwm()))
def pearson_correlation_coefficient(sites_a, sites_b):
"""PEarson correlation coefficient"""
def pcc(cola, colb):
cola_avg = sum(cola[l] for l in "ACTG") / 4.0
colb_avg = sum(colb[l] for l in "ACTG") / 4.0
return (sum(((cola[l]-cola_avg) * (colb[l]-colb_avg)) for l in "ACTG") /
math.sqrt(sum((cola[l]-cola_avg)**2 for l in "ACTG") *
sum((colb[l]-colb_avg)**2 for l in "ACTG")))
ma = bioutils.build_motif(sites_a)
mb = bioutils.build_motif(sites_b)
return sum(pcc(cola, colb) for (cola,colb) in zip(ma.pwm(), mb.pwm()))
def kullback_leibler_divergence(sites_a, sites_b):
"""Kullback-Leibler divergence between two sets of sites"""
def safe_log2(x):
return math.log(x,2) if x != 0 else 0.0
def kl(cola, colb):
return (sum(cola[l] * safe_log2(cola[l] / colb[l]) for l in "ACTG") +
sum(colb[l] * safe_log2(colb[l] / cola[l]) for l in "ACTG")) / 2.0
ma = bioutils.build_motif(sites_a)
mb = bioutils.build_motif(sites_b)
return sum(kl(cola, colb) for (cola,colb) in zip(ma.pwm(), mb.pwm()))
def pearson_correlation_coefficient(sites_a, sites_b):
"""PEarson correlation coefficient"""
def pcc(cola, colb):
cola_avg = sum(cola[l] for l in "ACTG") / 4.0
colb_avg = sum(colb[l] for l in "ACTG") / 4.0
return (sum(((cola[l] - cola_avg) * (colb[l] - colb_avg))
for l in "ACTG") / math.sqrt(
sum((cola[l] - cola_avg)**2 for l in "ACTG") * sum(
(colb[l] - colb_avg)**2 for l in "ACTG")))
ma = bioutils.build_motif(sites_a)
mb = bioutils.build_motif(sites_b)
return sum(pcc(cola, colb) for (cola, colb) in zip(ma.pwm(), mb.pwm()))
def kullback_leibler_divergence(sites_a, sites_b):
"""Kullback-Leibler divergence between two sets of sites"""
def safe_log2(x):
return math.log(x, 2) if x != 0 else 0.0
def kl(cola, colb):
return (sum(cola[l] * safe_log2(cola[l] / colb[l]) for l in "ACTG") +
sum(colb[l] * safe_log2(colb[l] / cola[l])
for l in "ACTG")) / 2.0
ma = bioutils.build_motif(sites_a)
mb = bioutils.build_motif(sites_b)
return sum(kl(cola, colb) for (cola, colb) in zip(ma.pwm(), mb.pwm()))
def average_log_likelihood_ratio(sites_a, sites_b):
"""Average Log-likelihood ratio distance"""
def safe_log2(x):
return math.log(x,2) if x != 0 else 0.0
def allr(cola, colb, cnta, cntb):
return (sum((cnta[l]*safe_log2(colb[l]/0.25) +
cntb[l]*safe_log2(cola[l]/0.25))
for l in "ACTG") /
sum(cnta[l] + cntb[l] for l in 'ACTG'))
ma = bioutils.build_motif(sites_a)
mb = bioutils.build_motif(sites_b)
# reformat biopython count matrices
counts_a = [dict((l, ma.counts[l][i]) for l in "ACTG") for i in xrange(ma.length)]
counts_b = [dict((l, mb.counts[l][i]) for l in "ACTG") for i in xrange(mb.length)]
return sum(allr(cola, colb, cnta, cntb)
for (cola,colb,cnta,cntb) in zip(ma.pwm(), mb.pwm(), counts_a, counts_b))
def average_log_likelihood_ratio(sites_a, sites_b):
"""Average Log-likelihood ratio distance"""
def safe_log2(x):
return math.log(x, 2) if x != 0 else 0.0
def allr(cola, colb, cnta, cntb):
return (sum((cnta[l] * safe_log2(colb[l] / 0.25) +
cntb[l] * safe_log2(cola[l] / 0.25))
for l in "ACTG") / sum(cnta[l] + cntb[l] for l in 'ACTG'))
ma = bioutils.build_motif(sites_a)
mb = bioutils.build_motif(sites_b)
# reformat biopython count matrices
counts_a = [
dict((l, ma.counts[l][i]) for l in "ACTG") for i in xrange(ma.length)
]
counts_b = [
dict((l, mb.counts[l][i]) for l in "ACTG") for i in xrange(mb.length)
]
return sum(
allr(cola, colb, cnta, cntb)
for (cola, colb, cnta,
cntb) in zip(ma.pwm(), mb.pwm(), counts_a, counts_b))
def export_PSFM(meta_sites, **kwargs):
"""Export Position-Specific-Frequency-Matrix"""
format = kwargs['format']
rows = export_base(meta_sites)
aligned = bioutils.run_lasagna([m[0].site_instance for m in meta_sites])
motif = bioutils.build_motif(aligned)
consensus = bioutils.degenerate_consensus(motif)
TF_name= ','.join(set(row['curation__TF__name'] for row in rows))
sp = ','.join(set('_'.join(row['site_instance__genome__organism'].split()) for row in rows))
lines = []
if format == 'JASPAR':
lines.append('> CollecTF_%s_%s' % (TF_name, sp))
lines.append('A [ %s ]' % (' '.join(map(str, motif.counts['A']))))
lines.append('C [ %s ]' % (' '.join(map(str, motif.counts['C']))))
lines.append('G [ %s ]' % (' '.join(map(str, motif.counts['G']))))
lines.append('T [ %s ]' % (' '.join(map(str, motif.counts['T']))))
elif format == 'TRANSFAC':
lines.append('ID %s' % TF_name)
lines.append('BF %s' % sp)
lines.append('PO\tA\tC\tG\tT')
lines.extend('%02d\t%d\t%d\t%d\t%d\t%s' % (po+1, motif.counts['A'][po],
motif.counts['C'][po],
motif.counts['G'][po],
motif.counts['T'][po],
consensus[po])
for po in range(motif.length))
lines.append('XX')
elif format == 'raw_fasta':
lines.append('>CollecTF_%s_%s' % (TF_name, sp))
lines.extend('%d\t%d\t%d\t%d' % (motif.counts['A'][po],
motif.counts['C'][po],
motif.counts['G'][po],
motif.counts['T'][po])
for po in range(motif.length))
return '\n'.join(lines)
def export_PSFM(meta_sites, **kwargs):
"""Export Position-Specific-Frequency-Matrix"""
format = kwargs['format']
rows = export_base(meta_sites)
aligned = bioutils.run_lasagna([m[0].site_instance for m in meta_sites])
motif = bioutils.build_motif(aligned)
consensus = bioutils.degenerate_consensus(motif)
TF_name = ','.join(set(row['curation__TF__name'] for row in rows))
sp = ','.join(
set('_'.join(row['site_instance__genome__organism'].split())
for row in rows))
lines = []
if format == 'JASPAR':
lines.append('> CollecTF_%s_%s' % (TF_name, sp))
lines.append('A [ %s ]' % (' '.join(map(str, motif.counts['A']))))
lines.append('C [ %s ]' % (' '.join(map(str, motif.counts['C']))))
lines.append('G [ %s ]' % (' '.join(map(str, motif.counts['G']))))
lines.append('T [ %s ]' % (' '.join(map(str, motif.counts['T']))))
elif format == 'TRANSFAC':
lines.append('ID %s' % TF_name)
lines.append('BF %s' % sp)
lines.append('PO\tA\tC\tG\tT')
lines.extend(
'%02d\t%d\t%d\t%d\t%d\t%s' %
(po + 1, motif.counts['A'][po], motif.counts['C'][po],
motif.counts['G'][po], motif.counts['T'][po], consensus[po])
for po in range(motif.length))
lines.append('XX')
elif format == 'raw_fasta':
lines.append('>CollecTF_%s_%s' % (TF_name, sp))
lines.extend('%d\t%d\t%d\t%d' %
(motif.counts['A'][po], motif.counts['C'][po],
motif.counts['G'][po], motif.counts['T'][po])
for po in range(motif.length))
return '\n'.join(lines)