def do_bootstrap():
"""
Do a bootstrap analysis on GO p-values.
"""
logging.info(
'Running GO bootstrap analysis with %d samples: topGO method=%s',
options.num_bootstrap_samples, options.topgo_method)
transcriptional_programs, factor_universe, target_universe = tp_threshold.threshold_tps(
)
genes_2_GO, go_context = go.initialise_go_context(factor_universe,
target_universe,
options.go_ontologies)
tp_sizes = filter(
None, map(len, (tp.targets for tp in transcriptional_programs)))
p_values = list()
for sample in generate_bootstrap_samples(options.num_bootstrap_samples,
target_universe, tp_sizes):
go_analysis = dict(
(ontology,
go.try_go_analysis(go_data, sample, 1., options.topgo_method))
for ontology, go_data in go_context.targets_go_data.iteritems())
best_p_value = min(
map(topgo.p_value_from_r,
map(topgo.best_p_value, go_analysis.values())))
p_values.append(best_p_value)
logging.info('GO bootstrap analysis completed')
return p_values
def symatlas():
logging.info('Analysing SymAtlas expression data.')
probes_to_genes = SA.probes_to_genes()
dataset, tissues, probes, fold_changes = SA.expression_data()
highly_expressed = SA.fold_change_above_median(fold_changes, fold_change=options.symatlas_fold_change_threshold)
highly_expressed_probes = SA.match_probe_sets(highly_expressed.T, probes)
highly_expressed_genes = [
set(probes_to_genes[p] for p in hep if p in probes_to_genes and probes_to_genes[p])
for hep in highly_expressed_probes
]
if False:
import pylab as P
P.figure()
P.bar(range(len(tissues)), map(len, highly_expressed_probes))
P.xlim(max=len(tissues))
transcriptional_programs, factor_universe, target_universe = tp_threshold.threshold_tps()
tester = gene_set_enrichment.TpEnrichmentTester(factor_universe, target_universe, p_value_threshold=1e-3)
for tissue, tissue_genes in zip(tissues, highly_expressed_genes):
logging.debug('Testing %d transcriptional programs\' factors for enrichment in genes over-expressed in tissue %s', len(transcriptional_programs), tissue)
for tp, (test_drawn, test_size, test_complement_size, draws, p_value) in tester.test_transcriptional_program_factors(transcriptional_programs, tissue_genes):
logging.info(
'TP:%4d; %4d in program; %4d/%5d over-expressed in % -32s; %4d in intersection; p-value=%e',
tp.k, draws, test_size, test_complement_size+test_size, tissue, test_drawn, p_value
)
for tp, (test_drawn, test_size, test_complement_size, draws, p_value) in tester.test_transcriptional_program_targets(transcriptional_programs, tissue_genes):
logging.info(
'TP:%4d; %4d in program; %4d/%5d over-expressed in % -32s; %4d in intersection; p-value=%g',
tp.k, draws, test_size, test_complement_size+test_size, tissue, test_drawn, p_value
)
def do_bootstrap():
"""
Do a bootstrap analysis on GO p-values.
"""
logging.info('Running GO bootstrap analysis with %d samples: topGO method=%s', options.num_bootstrap_samples, options.topgo_method)
transcriptional_programs, factor_universe, target_universe = tp_threshold.threshold_tps()
genes_2_GO, go_context = go.initialise_go_context(factor_universe, target_universe, options.go_ontologies)
tp_sizes = filter(None, map(len, (tp.targets for tp in transcriptional_programs)))
p_values = list()
for sample in generate_bootstrap_samples(options.num_bootstrap_samples, target_universe, tp_sizes):
go_analysis = dict(
(
ontology,
go.try_go_analysis(
go_data,
sample,
1.,
options.topgo_method
)
)
for ontology, go_data
in go_context.targets_go_data.iteritems()
)
best_p_value = min(map(topgo.p_value_from_r, map(topgo.best_p_value, go_analysis.values())))
p_values.append(best_p_value)
logging.info('GO bootstrap analysis completed')
return p_values
def validiate_program_28():
factor_validation_sets, target_validation_sets = generate_validation_sets()
transcriptional_programs, factor_universe, target_universe = threshold_tps()
factor_universe = set(factor_universe)
target_universe = set(target_universe)
restrict_validation_sets(factor_validation_sets.values(), factor_universe)
restrict_validation_sets(target_validation_sets.values(), target_universe)
tp = transcriptional_programs[28]
for name, reference_set in target_validation_sets.iteritems():
validate_tp_set(tp.k, "Factors", tp.targets, name, reference_set, target_universe)
def validiate_program_28():
factor_validation_sets, target_validation_sets = generate_validation_sets()
transcriptional_programs, factor_universe, target_universe = threshold_tps(
)
factor_universe = set(factor_universe)
target_universe = set(target_universe)
restrict_validation_sets(factor_validation_sets.values(), factor_universe)
restrict_validation_sets(target_validation_sets.values(), target_universe)
tp = transcriptional_programs[28]
for name, reference_set in target_validation_sets.iteritems():
validate_tp_set(tp.k, 'Factors', tp.targets, name, reference_set,
target_universe)
def validate_one_set(validation_sets, latex_f=None):
transcriptional_programs, factor_universe, target_universe = threshold_tps()
target_universe = set(target_universe)
restrict_validation_sets(validation_sets.values(), target_universe)
overall_p_values = list()
for tp in transcriptional_programs:
# logging.info('Validating transcriptional program %d against %d sets of targets.', tp.k, len(target_validation_sets))
tp_p_values = list()
for name, reference_set in validation_sets.iteritems():
validate_tp_set(tp.k, "Targets", tp.targets, name, reference_set, target_universe, latex_f, tp_p_values)
overall_p_values.append(
lou_jost_multiple_p_value_adjustment(reduce(float.__mul__, tp_p_values), len(tp_p_values))
)
return overall_p_values
def validate_one_set(validation_sets, latex_f=None):
transcriptional_programs, factor_universe, target_universe = threshold_tps(
)
target_universe = set(target_universe)
restrict_validation_sets(validation_sets.values(), target_universe)
overall_p_values = list()
for tp in transcriptional_programs:
#logging.info('Validating transcriptional program %d against %d sets of targets.', tp.k, len(target_validation_sets))
tp_p_values = list()
for name, reference_set in validation_sets.iteritems():
validate_tp_set(tp.k, 'Targets', tp.targets, name, reference_set,
target_universe, latex_f, tp_p_values)
overall_p_values.append(
lou_jost_multiple_p_value_adjustment(
reduce(float.__mul__, tp_p_values), len(tp_p_values)))
return overall_p_values
def validate():
"""
Validate the transcriptional programs against the validation sets.
"""
factor_validation_sets, target_validation_sets = generate_validation_sets()
transcriptional_programs, factor_universe, target_universe = threshold_tps()
factor_universe = set(factor_universe)
target_universe = set(target_universe)
restrict_validation_sets(factor_validation_sets.values(), factor_universe)
restrict_validation_sets(target_validation_sets.values(), target_universe)
latex_f = open(os.path.join(options.output_dir, "validation.tex"), "w")
for tp in transcriptional_programs:
# logging.info('Validating transcriptional program %d against %d sets of factors.', tp.k, len(factor_validation_sets))
for name, reference_set in factor_validation_sets.iteritems():
validate_tp_set(tp.k, "Factors", tp.factors, name, reference_set, factor_universe, latex_f)
# logging.info('Validating transcriptional program %d against %d sets of targets.', tp.k, len(target_validation_sets))
for name, reference_set in target_validation_sets.iteritems():
validate_tp_set(tp.k, "Targets", tp.targets, name, reference_set, target_universe, latex_f)
latex_f.close()
def go_analysis():
logging.info('Running GO analysis: p-value threshold=%e; topGO method=%s',
options.go_p_value_threshold, options.topgo_method)
transcriptional_programs, factor_universe, target_universe = tp_threshold.threshold_tps(
)
genes_2_GO, go_context = go.initialise_go_context(factor_universe,
target_universe,
options.go_ontologies)
go_analyses = list()
f = open(os.path.join(options.output_dir, 'go-analyses.txt'), 'w')
for tp in transcriptional_programs:
go_analysis = go.TPGoAnalysis(tp, go_context,
options.go_p_value_threshold,
options.topgo_method)
go_analysis.print_go_analyses(f,
options.go_p_value_threshold,
log=True)
go_analysis.print_go_analyses(sys.stdout, options.go_p_value_threshold)
go_analyses.append(go_analysis)
f.close()
write_latex(go_analyses)
def validate():
"""
Validate the transcriptional programs against the validation sets.
"""
factor_validation_sets, target_validation_sets = generate_validation_sets()
transcriptional_programs, factor_universe, target_universe = threshold_tps(
)
factor_universe = set(factor_universe)
target_universe = set(target_universe)
restrict_validation_sets(factor_validation_sets.values(), factor_universe)
restrict_validation_sets(target_validation_sets.values(), target_universe)
latex_f = open(os.path.join(options.output_dir, 'validation.tex'), 'w')
for tp in transcriptional_programs:
#logging.info('Validating transcriptional program %d against %d sets of factors.', tp.k, len(factor_validation_sets))
for name, reference_set in factor_validation_sets.iteritems():
validate_tp_set(tp.k, 'Factors', tp.factors, name, reference_set,
factor_universe, latex_f)
#logging.info('Validating transcriptional program %d against %d sets of targets.', tp.k, len(target_validation_sets))
for name, reference_set in target_validation_sets.iteritems():
validate_tp_set(tp.k, 'Targets', tp.targets, name, reference_set,
target_universe, latex_f)
latex_f.close()
def print_latex(go_analyses, f=sys.stdout, factors=True):
transcriptional_programs, factor_universe, target_universe = tp_threshold.threshold_tps(
)
t = factors and 'Factors' or 'Targets'
print >> f, '%% %s' % t
print >> f, 'TP & %s & GO term & & GO description & \\multicolumn{2}{c}{annotated} & $p$-score \\\\' % t
print >> f, '\\hline'
for k, (tp, tp_analysis) in enumerate(
zip(transcriptional_programs, go_analyses)):
if factors:
type_analysis = tp_analysis.factors_go_analysis
size = len(tp.factors)
else:
type_analysis = tp_analysis.targets_go_analysis
size = len(tp.targets)
for ontology, analysis in type_analysis.iteritems():
if None != analysis:
for go_id, go_term, annotated, significant, expected, pvalue in topgo.yield_stats(
analysis):
# Program & Factors & GO term & & GO description & \multicolumn{2}{c}{annotated} & $p$-score \\
print >> f, '% 4d & %5d & %s & %s & %42s & % 4d & %-4d & %.1e \\\\' % (
tp.k, size, go_id, ontology, go_term, significant,
annotated, pvalue)
print >> f, '\\\\'
