def load_seqs(filename):
"Load and convert sequences from fasta file."
logging.info('Loading sequences: %s', filename)
sequences = dict(sequences_from_fasta(filename))
numpy_seqs = dict((desc, hmm.preprocess_sequence(seq_to_numpy(seq))) for desc, (seq, tally) in sequences.iteritems())
tally = sum(imap(N.array, (tally for desc, (seq, tally) in sequences.iteritems())))
logging.info('Loaded %d sequences with %d bases', len(sequences), sum(imap(len, (seq for seq, tally in sequences.values()))))
return numpy_seqs, tally
def test_traits(self):
from hmm.pssm import create_background_model, PssmTraits, seq_to_numpy
from infpy import check_is_close_2
p_binding_site = .01
num_background_states = 2
emission_dists = [
[1., 0., 0., 0.],
[0., 1., 0., 0.],
[0., 1., 0., 0.],
[1., 0., 0., 0.],
[0., 0., 1., 0.],
[0., 0., 0., 1.],
[0., 0., 0., 1.],
[0., 0., 0., 1.],
[0., 0., 1., 0.],
[0., 1., 0., 0.],
[1., 0., 0., 0.],
[0., 1., 0., 0.],
[0., 0., 0., 1.],
]
K = len(emission_dists)
test_seq = 'accagtttgcact' # matches dist above
test_seq_order_0 = seq_to_numpy(test_seq)
# for various different orders
for order in [1, 2]:
# build a model of distribution above
traits = PssmTraits(K,
p_binding_site,
order,
num_background_states,
create_background_model,
emission_dists=emission_dists)
model = traits.new_model()
converted = hmm.model_states_2_model(model)
B = converted.B
# check the reverse complement states are correct
for n in xrange(model.N):
for o in xrange(model.M):
rev_comp_state, rev_comp_obs = traits.get_non_reverse_complement(
n, o)
assert check_is_close_2(
B[rev_comp_state, rev_comp_obs],
B[n, o]), ('%d,%d %d,%d: %f %f' %
(rev_comp_state, rev_comp_obs, n, o,
B[rev_comp_state, rev_comp_obs], B[n, o]))
# check viterbi gives correct result
test_seq_order_n = converted.converter.to_order_n(test_seq_order_0)
LL, states = converted.viterbi(test_seq_order_n)
for i, state in enumerate(states):
assert state == num_background_states + i
def evaluate_model(model, sequence):
"""
Evaluates the model against the sequence.
@return: True if there is at least one hit in the sequence
"""
hmm, traits = model
LL, states = hmm.viterbi(seq_to_numpy(sequence))
# we have a hit if we find at least K/2 states in the state sequence that are not in the
# background
return sum(state not in traits.background_states for state in states) > traits.K / 2
def load_seqs(filename):
"Load and convert sequences from fasta file."
logging.info('Loading sequences: %s', filename)
sequences = dict(sequences_from_fasta(filename))
numpy_seqs = dict((desc, hmm.preprocess_sequence(seq_to_numpy(seq)))
for desc, (seq, tally) in sequences.iteritems())
tally = sum(
imap(N.array, (tally for desc, (seq, tally) in sequences.iteritems())))
logging.info('Loaded %d sequences with %d bases', len(sequences),
sum(imap(len, (seq for seq, tally in sequences.values()))))
return numpy_seqs, tally
def evaluate_model(model, sequence):
"""
Evaluates the model against the sequence.
@return: True if there is at least one hit in the sequence
"""
hmm, traits = model
LL, states = hmm.viterbi(seq_to_numpy(sequence))
# we have a hit if we find at least K/2 states in the state sequence that are not in the
# background
return sum(state not in traits.background_states
for state in states) > traits.K / 2
def test_traits(self):
from hmm.pssm import create_background_model, PssmTraits, seq_to_numpy
from infpy import check_is_close_2
p_binding_site = .01
num_background_states = 2
emission_dists = [
[ 1., 0., 0., 0. ],
[ 0., 1., 0., 0. ],
[ 0., 1., 0., 0. ],
[ 1., 0., 0., 0. ],
[ 0., 0., 1., 0. ],
[ 0., 0., 0., 1. ],
[ 0., 0., 0., 1. ],
[ 0., 0., 0., 1. ],
[ 0., 0., 1., 0. ],
[ 0., 1., 0., 0. ],
[ 1., 0., 0., 0. ],
[ 0., 1., 0., 0. ],
[ 0., 0., 0., 1. ],
]
K = len(emission_dists)
test_seq = 'accagtttgcact' # matches dist above
test_seq_order_0 = seq_to_numpy(test_seq)
# for various different orders
for order in [1, 2]:
# build a model of distribution above
traits = PssmTraits(K, p_binding_site, order, num_background_states, create_background_model, emission_dists=emission_dists)
model = traits.new_model()
converted = hmm.model_states_2_model(model)
B = converted.B
# check the reverse complement states are correct
for n in xrange(model.N):
for o in xrange(model.M):
rev_comp_state, rev_comp_obs = traits.get_non_reverse_complement(n,o)
assert check_is_close_2(B[rev_comp_state,rev_comp_obs], B[n,o]), ('%d,%d %d,%d: %f %f' % (rev_comp_state,rev_comp_obs,n,o,B[rev_comp_state,rev_comp_obs],B[n,o]))
# check viterbi gives correct result
test_seq_order_n = converted.converter.to_order_n(test_seq_order_0)
LL, states = converted.viterbi(test_seq_order_n)
for i, state in enumerate(states):
assert state == num_background_states+i
'# site sequence;sequence;start;is reverse complement; state sequence\n'
)
for seq_idx, seq_sites in enumerate(sites):
for site, states, start, is_rev_comp in seq_sites:
if is_rev_comp:
site = hmm.pssm.rev_comp(site)
f.write('%s;%d;%d;%d;%s\n' %
(hmm.pssm.numpy_to_seq(site), seq_idx, start,
is_rev_comp, states))
if '__main__' == __name__:
from hmm.pssm import PssmTraits, create_background_model, seq_to_numpy, random_sequence, information_content
from random import random
site = seq_to_numpy('aaactcaa')
K = len(site)
rev_comp_site = seq_to_numpy('ttgagttt')
num_seqs = 60
seq_length = K + 30
start = 20
def gen_sequence():
seq = random_sequence(seq_length)
if random() > .5:
seq[start:start + K] = site
else:
seq[start:start + K] = rev_comp_site
return seq
p_binding_site = .01
option_parser.add_option(
"--threshold-graph",
dest="threshold_graph",
help="file to write an image showing how # seqs with site varies by threshold."
)
# sys.argv='dummy.py -m /home/reid/T00759.pssm -s /home/reid/T00759.fa --threshold-graph test.png'.split()
options, args = option_parser.parse_args()
for option in option_parser.option_list:
if option.dest:
logging.info('%s: %s (%s)', option.dest, str(getattr(options, option.dest)), option.help)
logging.info('Loading sequences: %s', options.sequences_file)
sequences = dict(sequences_from_fasta(options.sequences_file))
numpy_seqs = dict((desc, seq_to_numpy(seq)) for desc, seq in sequences.iteritems())
logging.info('Loaded %d sequences', len(sequences))
logging.info('Parsing PSSMs: %s', options.models_file)
pssms = list(parse_models(open(options.models_file)))
logging.info('Building models')
models = [
build_hmm_from_semi_parsed(parsed, p_binding_site=options.p_binding_site)
for parsed in pssms
]
logging.info('Analysing sequences')
p_binding_sites = list()
f.write('# site sequence;sequence;start;is reverse complement; state sequence\n')
for seq_idx, seq_sites in enumerate(sites):
for site, states, start, is_rev_comp in seq_sites:
if is_rev_comp:
site = hmm.pssm.rev_comp(site)
f.write('%s;%d;%d;%d;%s\n' % (hmm.pssm.numpy_to_seq(site), seq_idx, start, is_rev_comp, states))
if '__main__' == __name__:
from hmm.pssm import PssmTraits, create_background_model, seq_to_numpy, random_sequence, information_content
from random import random
site = seq_to_numpy('aaactcaa')
K = len(site)
rev_comp_site = seq_to_numpy('ttgagttt')
num_seqs = 60
seq_length = K + 30
start = 20
def gen_sequence():
seq = random_sequence(seq_length)
if random() > .5:
seq[start:start+K] = site
else:
seq[start:start+K] = rev_comp_site
return seq
p_binding_site = .01
order = 1