def test_pssm_pseudo_counts():
sascha_pssms = biopsy.SequenceVec()
sascha_acc = 'M00975'
# sascha_seq = 'gtaaaccaggctgcctGAgaacttgttgcgaatcc'
sascha_seq = 'ttgttgcga'
sascha_seq = 'ttgttgcaa'
# plot_likelihoods( biopsy.get_pssm( 'M00975' ), 'M00975' )
# plot_likelihoods( biopsy.get_pssm( 'R02146' ), 'R02146' )
print 'Binding,Background,odds,p(binding),cumulative p(binding),Sequence'
biopsy.PssmParameters.singleton().use_p_value = True;
# biopsy.PssmParameters.singleton().binding_background_odds_prior = 1;
for pc in [ 0.0, 0.25, 0.5, 1.0, 2.0 ]:
# force cache load
biopsy.get_pssm( sascha_acc )
biopsy.clear_pssm_cache()
biopsy.PssmParameters.singleton().pseudo_counts = pc
p = biopsy.get_pssm( sascha_acc )
score = biopsy.score_pssm( p.pssm, sascha_seq )
(
bind,
back,
cum_bind,
cum_back,
odds_ratio,
cum_odds_ratio,
p_bind,
cum_p_bind,
p_value_p_bind
) = biopsy.get_pssm_likelihoods_for_score( p, score )
print pc,
print \
'%f,%f,%f,%f,%f,%f,%f' \
% \
( bind, back, cum_bind, cum_back, p_bind, cum_p_bind, p_value_p_bind )
biopsy.plot_likelihoods( p, sascha_acc + ': ' + str( pc ), score )
# print 'Trying with standard distributions'
# biopsy.PssmParameters.singleton().use_cumulative_dists = False;
# hits = biopsy.HitVec()
# biopsy.score_pssm_on_sequence( sascha_acc, sascha_seq, 0.001, hits )
# print hits
print 'Trying with cumulative distributions'
biopsy.PssmParameters.singleton().use_cumulative_dists = True;
hits = biopsy.HitVec()
biopsy.score_pssm_on_sequence( sascha_acc, sascha_seq, 0.001, hits )
print hits
print
def test_pssm_score():
# 'V$AP1_Q2'
pssm_acc = biopsy.get_transfac_pssm_accession( 'V$DEAF1_01' );
pssm_info = biopsy.get_pssm( pssm_acc )
# print pssm_info.pssm
seq = 'tacatcatctgtctgcagtagtctaaccgaccccccccagttttagaagcagactgcatgcggacgggaccgcggatcgcgcggtgcgcctcagtgtacttccgaacgaatgagtcattaatagagcgctatatcgtaactgtctttgacgaagtataccgaaaccgtgcagccagacgtgatccgggcgttgtaaaggcgatcagcgccctaggagtaccatttttgccgtaggcttgcgtctcaaagaccagctggggcgtggtatcactcgtcagtacgatttctgccagatagatagcatagactgaaccttaggcccaatagggacacaattacccgagtgactgactggtctaaggggagtccccccttaaaacgttttacgtaatagcgggctccagaagcaaagcatcggtttgagccccagtactaaacgtttgagtgtttgctctcgtctgataggtaaaccgacaagagaaccaagctcaaggcgcggtaggtgcgccttgcgaactgttgatgccgtgagcgccaccatcccgtgcatcataggcagggagagaagaccacatggccttgcgaccgtatgagctgtttcagattaaatgccaacgggcatggtcggtgtccagcattttttgcagtcagctggtggtacacagtggggacaagaacgcctctggtagatgtcttctgaaggagtaactcatttcgttgaatcgaccttcccttgcgcttgaacgcggacctctagtctctctcgcagactggggtcgaaaatcaaggtagatatggaatgttccgcatgagggtagcgaccggatcgggcgtcaagtatatcctccctgctacgtccccctactagcctcagtccgcctcgaacctaggaagattggccacatcagcttggtggatgcctggtccatacttcagacccgagaatgttagacaggaccccatttggctcctttacgtacgatctatgtagacgcagtga'
for i in range( len( seq ) - len( pssm_info.pssm ) + 1 ):
s = biopsy.score_pssm( pssm_info.pssm, seq[i:] )
p_binding = biopsy.get_p_binding(
biopsy.get_odds_ratio(
s,
pssm_info.get_dist( True, False ),
pssm_info.get_dist( False, False ) ) )
if p_binding > 0.05:
print i, s, p_binding
result = biopsy.HitVec()
p_binding = biopsy.score_pssm_on_sequence( pssm_acc, seq, 0.05, result )
print 'Got', len( result ), 'hits from', len( seq ), 'bases'
print p_binding
for species in remo.get_sequence_ids():
yield remo.get_sequence_for(species, True)
def histogram(acc, score_counts):
import pylab, numpy
pylab.clf()
pylab.bar(xrange(num_buckets), numpy.power(score_counts, 0.25))
pylab.savefig('graphs/%s.png' % acc)
try:
remome
except NameError:
remome_file = 'c:/data/remos/100/100.filtered'
print 'Loading remome: %s' % remome_file
remome = biopsy.Remome.load(remome_file)
score_counts = {}
for acc in itertools.islice(pssm_accs(), num_pssms):
score_counts[acc] = numpy.zeros(num_buckets, numpy.uint8)
for seq in itertools.islice(sequences_from_remome(remome), num_sequences):
hits = biopsy.HitVec()
p_bind = biopsy.score_pssm_on_sequence(pssm_name=acc,
threshold=0.0,
sequence=seq,
result=hits)
for h in hits:
score_counts[acc][int(h.p_binding * num_buckets)] += 1
histogram(acc, score_counts[acc])
for aligned in remome.get_aligned_sequences():
for remo in remome.get_remos_for(aligned):
for species in remo.get_sequence_ids():
yield remo.get_sequence_for(species, True)
def histogram(acc, score_counts):
import pylab, numpy
pylab.clf()
pylab.bar(xrange(num_buckets), numpy.power(score_counts, 0.25))
pylab.savefig("graphs/%s.png" % acc)
try:
remome
except NameError:
remome_file = "c:/data/remos/100/100.filtered"
print "Loading remome: %s" % remome_file
remome = biopsy.Remome.load(remome_file)
score_counts = {}
for acc in itertools.islice(pssm_accs(), num_pssms):
score_counts[acc] = numpy.zeros(num_buckets, numpy.uint8)
for seq in itertools.islice(sequences_from_remome(remome), num_sequences):
hits = biopsy.HitVec()
p_bind = biopsy.score_pssm_on_sequence(pssm_name=acc, threshold=0.0, sequence=seq, result=hits)
for h in hits:
score_counts[acc][int(h.p_binding * num_buckets)] += 1
histogram(acc, score_counts[acc])
