def test_get_default_network_scaling(self):
scaling_fun = scoring.get_default_network_scaling(100)
self.assertAlmostEqual(0.0, scaling_fun(1))
self.assertAlmostEqual(0.013513514, scaling_fun(3))
self.assertAlmostEqual(0.493243243, scaling_fun(74))
self.assertAlmostEqual(0.5, scaling_fun(75))
self.assertAlmostEqual(0.5, scaling_fun(76))
self.assertAlmostEqual(0.5, scaling_fun(100))
def test_net_scoring(self):
#tests the network scoring by itself#
network_scaling_fun = scoring.get_default_network_scaling(2000)
network_scoring = nw.ScoringFunction(self.organism,
self.membership,
self.ratio_matrix,
scaling_func=network_scaling_fun,
run_in_iteration=lambda x: True,
config_params=self.config_params)
netscores = network_scoring.compute(self.iteration_result).sorted_by_row_name()
ref_netscores = read_matrix('testdata/netscores_fixed.tsv')
self.assertTrue(check_matrix_values(netscores, ref_netscores))
def make_row_scoring(self):
"""makes a row scoring function on demand"""
# Default row scoring functions
row_scoring = microarray.RowScoringFunction(
self.membership(), self.ratio_matrix,
scaling_func=lambda iteration: self['row_scaling'],
config_params=self.config_params)
self.row_scoring = row_scoring
meme_suite = meme.MemeSuite430()
sequence_filters = [
motif.unique_filter,
motif.get_remove_low_complexity_filter(meme_suite),
motif.get_remove_atgs_filter(self['search_distances']['upstream'])]
motif_scaling_fun = scoring.get_default_motif_scaling(
self['num_iterations'])
motif_scoring = motif.MemeScoringFunction(
self.organism(),
self.membership(),
self.ratio_matrix,
meme_suite,
sequence_filters=sequence_filters,
scaling_func=motif_scaling_fun,
num_motif_func=motif.default_nmotif_fun,
update_in_iteration=scoring.schedule(601, 3),
motif_in_iteration=scoring.schedule(600, 100),
#update_in_iteration=scoring.schedule(100, 10),
#motif_in_iteration=scoring.schedule(100, 100),
config_params=self.config_params)
self.motif_scoring = motif_scoring
network_scaling_fun = scoring.get_default_network_scaling(
self['num_iterations'])
network_scoring = nw.ScoringFunction(
self.organism(),
self.membership(),
self.ratio_matrix,
scaling_func=network_scaling_fun,
run_in_iteration=scoring.schedule(1, 7),
config_params=self.config_params)
self.network_scoring = network_scoring
row_scoring_functions = [row_scoring, motif_scoring, network_scoring]
return scoring.ScoringFunctionCombiner(
self.membership(),
row_scoring_functions,
config_params=self.config_params,
log_subresults=True)
def make_row_scoring(self):
"""returns the row scoring function"""
# we need to remove the location from the sequence when selecting for
# individual clusters
sequence_filters = [lambda seqs, feature_ids: {key: seqs[key][1] for key in seqs.keys()}]
network_scaling_fun = scoring.get_default_network_scaling(self['num_iterations'])
meme_scoring = None
weeder_scoring = None
row_scoring = microarray.RowScoringFunction(
self.membership(), self.ratio_matrix,
lambda iteration: ROW_WEIGHT,
config_params=self.config_params)
scoring_funcs = [row_scoring,
self.make_network_scoring(network_scaling_fun)]
if ADD_SET_ENRICHMENT:
scoring_funcs.append(self.make_set_scoring(network_scaling_fun,
USE_SET_TYPES))
if ADD_MEME:
meme_suite_meme = self.meme_suite('upstream')
meme_scoring = self.make_meme_scoring('upstream',
meme_suite_meme,
sequence_filters + [motif.get_remove_low_complexity_filter(meme_suite_meme)])
if ADD_WEEDER:
meme_suite_weeder = self.meme_suite(WEEDER_SEQ_TYPE)
weeder_scoring = self.make_weeder_scoring(WEEDER_SEQ_TYPE,
meme_suite_weeder,
sequence_filters + [motif.get_remove_low_complexity_filter(meme_suite_weeder)])
if ADD_MEME and ADD_WEEDER:
scoring_funcs.append(scoring.ScoringFunctionCombiner(
self.membership(),
[meme_scoring, weeder_scoring],
scaling_func=lambda iteration: 0.5,
config_params=self.config_params))
else:
if ADD_MEME:
scoring_funcs.append(meme_scoring)
if ADD_WEEDER:
scoring_funcs.append(weeder_scoring)
return scoring.ScoringFunctionCombiner(self.membership(), scoring_funcs,
config_params=self.config_params)
def make_row_scoring(self):
"""returns the row scoring function"""
sequence_filters = []
network_scaling_fun = scoring.get_default_network_scaling(self['num_iterations'])
meme_scoring = None
weeder_scoring = None
row_scoring = microarray.RowScoringFunction(
self.membership(), self.ratio_matrix,
lambda iteration: self['row_scaling'],
config_params=self.config_params)
scoring_funcs = [row_scoring,
self.make_network_scoring(network_scaling_fun)]
if ADD_MEME:
meme_scoring = self.make_meme_scoring('upstream',
self.meme_suite('upstream'),
sequence_filters)
if ADD_WEEDER:
weeder_scoring = self.make_weeder_scoring(WEEDER_SEQ_TYPE,
self.meme_suite(WEEDER_SEQ_TYPE),
sequence_filters)
if ADD_MEME and ADD_WEEDER:
scoring_funcs.append(scoring.ScoringFunctionCombiner(
self.membership(),
[meme_scoring, weeder_scoring],
scaling_func=lambda iteration: 0.5,
config_params=self.config_params))
else:
if ADD_MEME:
scoring_funcs.append(meme_scoring)
if ADD_WEEDER:
scoring_funcs.append(weeder_scoring)
return scoring.ScoringFunctionCombiner(self.membership(), scoring_funcs,
config_params=self.config_params)
def __init__(self):
scaling_fun = scoring.get_default_network_scaling(2000)
scoring.ScoringFunctionBase.__init__(self, None, None, scaling_fun)
def make_row_scoring(self):
"""returns the row scoring function"""
row_scoring = microarray.RowScoringFunction(
self.membership(), self.ratio_matrix,
scaling_func=lambda iteration: self['row_scaling'],
config_params=self.config_params)
background_file_prom = meme_FS.global_background_file_FS(
self.organism(), self.ratio_matrix.row_names(), 'Promoter',
alphabet, alphabet_replacement,
use_revcomp=True, resultsDir = "/".join([self['meme_dir'], 'backgroundFiles']))
meme_suite = meme_FS.MemeSuite481(background_file=background_file_prom,
remove_tempfiles=True,
resultsDir = self['meme_dir'])
"""
background_file_p3utr = meme_FS.global_background_file_FS(
self.organism(), self.ratio_matrix.row_names(), '3pUTR',
alphabet, alphabet_replacement,
use_revcomp=True)
meme_suite_p3utr = meme_FS.MemeSuite430(background_file=background_file_p3utr)
"""
sequence_filters = [
motif_FS.unique_filter,
motif_FS.get_remove_low_complexity_filter(meme_suite),
motif_FS.get_remove_atgs_filter(self['search_distances']['Promoter'])]
sequence_filters = [motif_FS.get_remove_atgs_filter(self['search_distances']['Promoter'])]
motif_scaling_fun = scoring.get_default_motif_scaling(self['num_iterations'])
motif_scoring = motif_FS.MemeScoringFunction(
self.organism(),
self.membership(),
self.ratio_matrix,
meme_suite,
seqtype='Promoter',
sequence_filters=sequence_filters,
pvalue_filter=motif_FS.MinPValueFilter(-9),
scaling_func=motif_scaling_fun,
update_in_iteration=scoring.schedule(11, 3),
motif_in_iteration=scoring.schedule(10, 100),
config_params=self.config_params)
"""
motif_scoring_3pUTR = motif_FS.MemeScoringFunction(
self.organism(),
self.membership(),
self.ratio_matrix,
meme_suite,
seqtype='3pUTR',
sequence_filters=sequence_filters,
pvalue_filter=motif_FS.MinPValueFilter(-9),
scaling_func=motif_scaling_fun,
run_in_iteration=scoring.schedule(100, 50),
config_params=self.config_params)
motif_combiner = scoring.ScoringFunctionCombiner(
self.membership(),
[motif_scoring, motif_scoring_3pUTR],
scaling_func=lambda iteration: 0.5)
"""
'''
weeder_scoring = motif.WeederScoringFunction(
self.organism(), self.membership(), self.ratio_matrix,
meme_suite_p3utr, '3pUTR',
pvalue_filter=motif.MinPValueFilter(-20.0),
scaling_func=lambda iteration: 0.0,
run_in_iteration=scoring.default_motif_iterations,
config_params=self.config_params)
motif_combiner = scoring.ScoringFunctionCombiner(
self.membership(),
[motif_scoring, weeder_scoring],
scaling_func=lambda iteration: 0.5)
'''
network_scaling_fun = scoring.get_default_network_scaling(self['num_iterations'])
network_scoring = nw.ScoringFunction(self.organism(),
self.membership(),
self.ratio_matrix,
scaling_func=network_scaling_fun,
run_in_iteration=scoring.schedule(1, 7),
config_params=self.config_params)
row_scoring_functions = [row_scoring, motif_scoring, network_scoring]
#row_scoring_functions = [row_scoring]
return scoring.ScoringFunctionCombiner(self.membership(),
row_scoring_functions,
log_subresults=True)
