def show_loo_prediction():
print 'Drawing LOO prediction and error...'
iLOO = 18
train_mask = np.arange(len(x)) != iLOO
x_train = x[train_mask]
y_train = y[train_mask]
theta,_,_,_ = fitter.fit(x_train,y_train)
fig = plot_one_series(series,shape,theta,yrange,train_mask=train_mask)
save_figure(fig,'RP/methods-3-LOO-prediction.png', under_results=True)
def show_loo_prediction():
print 'Drawing LOO prediction and error...'
iLOO = 18
train_mask = np.arange(len(x)) != iLOO
x_train = x[train_mask]
y_train = y[train_mask]
theta, _, _, _ = fitter.fit(x_train, y_train)
fig = plot_one_series(series, shape, theta, yrange, train_mask=train_mask)
save_figure(fig, 'RP/methods-3-LOO-prediction.png', under_results=True)
def show_loo_score():
print 'Drawing LOO prediction for all points and R2 score...'
theta, _, test_preds, _ = fitter.fit(x, y, loo=True)
fig = plot_one_series(series,
shape,
theta=None,
yrange=yrange,
test_preds=test_preds)
save_figure(fig, 'RP/methods-4-R2-score.png', under_results=True)
def do_gene_fits(data, gene, fitter, filename, b_show):
fig = plot_gene(data,gene)
if filename is None:
ensure_dir(results_dir())
filename = join(results_dir(), 'fit.png')
print 'Saving figure to {}'.format(filename)
save_figure(fig, filename)
if b_show:
plt.show(block=True)
def do_one_fit(series, fitter, loo, filename, b_show):
if fitter is not None:
theta, sigma, LOO_predictions,_ = fitter.fit(series.ages, series.single_expression, loo=loo)
fig = plot_one_series(series, fitter.shape, theta, LOO_predictions)
else:
fig = plot_one_series(series)
if filename is None:
ensure_dir(results_dir())
filename = join(results_dir(), 'fit.png')
save_figure(fig, filename, print_filename=True)
if b_show:
plt.show(block=True)
def analyze_pathway(pathway, data, fitter, fits, html_only=False):
print 80 * '='
print 'Analyzing pathway {}'.format(pathway)
print 80 * '='
dct_tuples = {}
for region in data.region_names:
dct_region_tuples, region_correlations = analyze_one_region(
data, fitter, fits, region)
fig = plot_gene_correlations_single_region(region_correlations, region,
data.gene_names)
filename = join(
'RP', 'correlation-heat-map-{}-{}.png'.format(region, pathway))
save_figure(fig, filename, b_close=True, under_results=True)
dct_tuples.update(dct_region_tuples)
print_best_improvements(dct_tuples)
tuples = dct_tuples.values()
pairs = [(x[0], x[1]) for x in tuples]
fig = plot_comparison_scatter(pairs, pathway)
filename = join('RP', 'correlation-diff-scatter-{}.png'.format(pathway))
save_figure(fig, filename, b_close=True, under_results=True)
fig = plot_comparison_bar(tuples)
filename = join('RP', 'correlation-diff-bar-{}.png'.format(pathway))
save_figure(fig, filename, b_close=True, under_results=True)
fig = plot_comparison_bar(tuples, several_levels=True)
filename = join(
'RP', 'correlation-diff-bar-several-levels-{}.png'.format(pathway))
save_figure(fig, filename, b_close=True, under_results=True)
def fit_serveral_genes(series, fitter, loo, filename, b_show):
if fitter is not None:
theta, L, LOO_predictions,_ = fitter.fit(series.ages, series.expression, loo=loo)
print 'L = {}'.format(L)
fig = plot_series(series, fitter.shape, theta, LOO_predictions)
else:
fig = plot_series(series)
if filename is None:
ensure_dir(results_dir())
filename = join(results_dir(), 'fits.png')
print 'Saving figure to {}'.format(filename)
save_figure(fig, filename)
if b_show:
plt.show(block=True)
def analyze_pathway(pathway, data, fitter, fits, html_only=False):
print 80 * '='
print 'Analyzing pathway {}'.format(pathway)
print 80 * '='
dct_tuples = {}
for region in data.region_names:
dct_region_tuples, region_correlations = analyze_one_region(data, fitter, fits, region)
fig = plot_gene_correlations_single_region(region_correlations, region, data.gene_names)
filename = join('RP','correlation-heat-map-{}-{}.png'.format(region,pathway))
save_figure(fig, filename, b_close=True, under_results=True)
dct_tuples.update(dct_region_tuples)
print_best_improvements(dct_tuples)
tuples = dct_tuples.values()
pairs = [(x[0],x[1]) for x in tuples]
fig = plot_comparison_scatter(pairs,pathway)
filename = join('RP','correlation-diff-scatter-{}.png'.format(pathway))
save_figure(fig, filename, b_close=True, under_results=True)
fig = plot_comparison_bar(tuples)
filename = join('RP','correlation-diff-bar-{}.png'.format(pathway))
save_figure(fig, filename, b_close=True, under_results=True)
fig = plot_comparison_bar(tuples, several_levels=True)
filename = join('RP','correlation-diff-bar-several-levels-{}.png'.format(pathway))
save_figure(fig, filename, b_close=True, under_results=True)
def plot_means(dataset):
min_age = min(dataset.ages)
max_age = max(dataset.ages)
min_expression = np.nanmin(dataset.expression.flat)
max_expression = np.nanmax(dataset.expression.flat)
center = np.empty(dataset.ages.shape)
std_plus = np.empty(dataset.ages.shape)
std_minus = np.empty(dataset.ages.shape)
for i, age in enumerate(dataset.ages):
a = dataset.expression[i, :, :].flat
c = nanmean(a)
s = nanstd(a)
center[i] = c
std_plus[i] = c + s
std_minus[i] = c - s
fig = plt.figure()
ax = fig.add_axes([0.08, 0.15, 0.85, 0.8])
ax.set_ylabel('expression level', fontsize=cfg.fontsize)
ax.set_xlabel('age', fontsize=cfg.fontsize)
ax.set_title('Mean expression across all genes - {}'.format(dataset.name),
fontsize=cfg.fontsize)
# set the development stages as x labels
stages = [stage.scaled(scaler) for stage in dev_stages]
ax.set_xticks([stage.central_age for stage in stages])
ax.set_xticklabels([stage.short_name for stage in stages],
fontsize=cfg.xtick_fontsize,
fontstretch='condensed',
rotation=90)
ax.set_xlim([min_age, max_age])
# mark birth time with a vertical line
ymin, ymax = ax.get_ylim()
birth_age = scaler.scale(0)
ax.plot([birth_age, birth_age], [ymin, ymax], '--', color='0.85')
ax.plot([min_age, max_age], [min_expression, min_expression], '--g')
ax.plot([min_age, max_age], [max_expression, max_expression], '--g')
ax.plot(dataset.ages, center, 'bx')
ax.plot(dataset.ages, std_plus, 'g-')
ax.plot(dataset.ages, std_minus, 'g-')
save_figure(fig,
'mean-expression-{}.png'.format(dataset.name),
under_results=True)
def fit_serveral_genes(series, fitter, loo, filename, b_show):
if fitter is not None:
theta, L, LOO_predictions, _ = fitter.fit(series.ages,
series.expression,
loo=loo)
print 'L = {}'.format(L)
fig = plot_series(series, fitter.shape, theta, LOO_predictions)
else:
fig = plot_series(series)
if filename is None:
ensure_dir(results_dir())
filename = join(results_dir(), 'fits.png')
print 'Saving figure to {}'.format(filename)
save_figure(fig, filename)
if b_show:
plt.show(block=True)
def plot_means(dataset):
min_age = min(dataset.ages)
max_age = max(dataset.ages)
min_expression = np.nanmin(dataset.expression.flat)
max_expression = np.nanmax(dataset.expression.flat)
center = np.empty(dataset.ages.shape)
std_plus = np.empty(dataset.ages.shape)
std_minus = np.empty(dataset.ages.shape)
for i, age in enumerate(dataset.ages):
a = dataset.expression[i, :, :].flat
c = nanmean(a)
s = nanstd(a)
center[i] = c
std_plus[i] = c + s
std_minus[i] = c - s
fig = plt.figure()
ax = fig.add_axes([0.08, 0.15, 0.85, 0.8])
ax.set_ylabel("expression level", fontsize=cfg.fontsize)
ax.set_xlabel("age", fontsize=cfg.fontsize)
ax.set_title("Mean expression across all genes - {}".format(dataset.name), fontsize=cfg.fontsize)
# set the development stages as x labels
stages = [stage.scaled(scaler) for stage in dev_stages]
ax.set_xticks([stage.central_age for stage in stages])
ax.set_xticklabels(
[stage.short_name for stage in stages], fontsize=cfg.xtick_fontsize, fontstretch="condensed", rotation=90
)
ax.set_xlim([min_age, max_age])
# mark birth time with a vertical line
ymin, ymax = ax.get_ylim()
birth_age = scaler.scale(0)
ax.plot([birth_age, birth_age], [ymin, ymax], "--", color="0.85")
ax.plot([min_age, max_age], [min_expression, min_expression], "--g")
ax.plot([min_age, max_age], [max_expression, max_expression], "--g")
ax.plot(dataset.ages, center, "bx")
ax.plot(dataset.ages, std_plus, "g-")
ax.plot(dataset.ages, std_minus, "g-")
save_figure(fig, "mean-expression-{}.png".format(dataset.name), under_results=True)
def run(self,duration):
start = datetime.now()
self.env.state = self.child.run()
self.agent.on()
if self.plot: save_figure(self.env.x,self.env.y,self.env.state,[self.agent.position[0]],[self.agent.position[1]])
while self.agent.active:
self.agent.precepts(self.env.state)
self.env.state = self.agent.effectors(self.env.state)
if self.plot: save_figure(self.env.x,self.env.y,self.env.state,[self.agent.position[0]],[self.agent.position[1]])
else: sleep(0.5)
if (datetime.now() - start).seconds >= duration:
self.agent.gohome()
self.agent.precepts(self.env.state)
self.env.state = self.agent.effectors(self.env.state)
if self.plot: save_figure(self.env.x,self.env.y,self.env.state,[self.agent.position[0]],[self.agent.position[1]])
else: sleep(0.5)
rwd, clnd = self.agent.reward, self.agent.cleaned
self.agent.reset()
return rwd, clnd
fitters = [Fitter(shape, sigma_prior='normal') for shape in shapes]
#####################################################
# Example fits
#####################################################
GRs = [
('ABHD4', 'STC', (5, 8)),
]
for g, r, yrange in GRs:
for fitter in fitters:
print 'Doing {}@{}...'.format(g, r)
series = data.get_one_series(g, r)
theta, _, _, _ = fitter.fit(series.ages, series.single_expression)
fig = plot_one_series(series, fitter.shape, theta, yrange)
save_figure(fig,
'RP/fit-examples-{}-{}-{}.png'.format(
fitter.shape.cache_name(), g, r),
under_results=True)
#####################################################
# Comparison for whole pathway
#####################################################
pathway = '17full'
data = data.restrict_pathway(pathway)
fits = [
get_all_fits(data, fitter, allow_new_computation=False)
for fitter in fitters
]
fig = plot_comparison_bar(data, shapes, fits)
save_figure(fig,
'RP/sigslope-comparison-bar-{}.png'.format(data.pathway),
('HTR1A', 'MFC'),
]
n_bins = 50
n_samples = 10
disable_all_warnings()
cfg.verbosity = 1
age_scaler = LogScaler()
data = GeneData.load('both').restrict_pathway(pathway).scale_ages(age_scaler)
shape = Sigmoid(priors='sigmoid_wide')
fitter = Fitter(shape, sigma_prior='normal')
fits = get_all_fits(data, fitter, allow_new_computation=False)
dirname = 'bootstrap'
fits = add_change_distributions(data, fitter, fits, n_bins=n_bins)
fig = plot_bootstrap_onset_variance(data, fits)
save_figure(fig, '{}/onset-variance-{}.png'.format(dirname, pathway), under_results=True, b_close=True)
fig = plot_change_width_scatter(data, fitter, fits)
save_figure(fig, '{}/width-scatter-{}.png'.format(dirname, pathway), under_results=True, b_close=True)
for g,r in gene_regions:
ds_name = data.region_to_dataset()[r]
fit = fits[ds_name][(g,r)]
fig = plot_bootstrap_fits(data, fit, n_bins=n_bins, n_samples=n_samples)
save_figure(fig, '{}/fits-{}-{}.png'.format(dirname,g,r), under_results=True, b_close=True)
fig = plot_bootstrap_histograms(data, fit, n_bins=n_bins, n_samples=n_samples)
save_figure(fig, '{}/transition-distribution-{}-{}.png'.format(dirname,g,r), under_results=True, b_close=True)
def basic_fit():
print 'Drawing basic fit...'
theta, _, _, _ = fitter.fit(x, y)
fig = plot_one_series(series, shape, theta, yrange)
save_figure(fig, 'RP/methods-1-basic-fit.png', under_results=True)
r,pval = spearmanr(onset_times, range(len(regions)))
return r,pval,lst_R2
lst_pathways = [
'serotonin',
'dopamine',
]
for pathway in lst_pathways:
data = GeneData.load('both').restrict_pathway(pathway).scale_ages(age_scaler)
shape = Sigmoid(priors='sigmoid_wide')
fitter = Fitter(shape, sigma_prior='normal')
fits = get_all_fits(data, fitter, allow_new_computation=False)
# R2_threshold = 0.5 YYY problem - we might be using bad fits.
regions = ['OFC', 'M1C', 'S1C', 'IPC', 'V1C']
scores = []
for g in data.gene_names:
r,pval,lst_R2 = get_gene_correlation(fits,g,regions)
scores.append( (g,r,pval,lst_R2) )
fig = plot_correlation_histogram(scores,pathway)
save_figure(fig,'{}/gradual-maturation-hist.png'.format(pathway,pathway), under_results=True, b_close=True)
for fR2 in [np.mean]: #[min,max,np.mean]:
fig = plot_scatter(scores, pathway, fR2)
save_figure(fig,'{}/gradual-maturation-scatter-{}.png'.format(pathway,fR2.__name__), under_results=True, b_close=True)
create_top_correlations_html(data,fitter,fits,scores,regions)
rho, pval = paired_spearman(pathway_mu)
scores.append( (-np.log10(pval), pval, rho, pathway) )
scores.sort(reverse=True)
save_scores(singles, scores, order)
##############################################################
# main
##############################################################
if __name__ == '__main__':
cfg.verbosity = 1
parser = argparse.ArgumentParser()
parser.add_argument('--list', help='Pathways list name. Default=brain_go_num_genes_min_15', default='brain_go_num_genes_min_15', choices=['all'] + pathway_lists.all_pathway_lists())
parser.add_argument('--cortex_only', help='Use only cortical regions', action='store_true')
parser.add_argument('--draw', help='Draw plot for this pathway and exit')
args = parser.parse_args()
if args.cortex_only:
order = 'V1C A1C S1C M1C DFC MFC OFC'.split()
else:
order = 'MD STR V1C OFC'.split()
singles = SingleRegion(args.list)
if args.draw is None:
timing_vs_region_order(singles, order)
else:
pathway = args.draw
fig = plot_pathway(singles, pathway, order)
filename = 'spearman-{}.png'.format(pathway)
save_figure(fig, filename, under_results=True)
cfg.verbosity = 1
age_scaler = LogScaler()
data = GeneData.load('both').scale_ages(age_scaler)
shapes = [
Sigmoid('sigmoid_wide'),
Poly(1, 'poly1'),
Poly(3, 'poly3'),
Spline()
]
GRs = [
('ADRB1', 'A1C', (5, 8)),
('GLRA2', 'STC', (5, 12)),
('TUBA1A', 'V1C', (10, 14)),
]
for g, r, yrange in GRs:
print 'Doing {}@{}...'.format(g, r)
thetas = []
for shape in shapes:
series = data.get_one_series(g, r)
sigma_prior = 'normal' if not isinstance(shape, Spline) else None
fitter = Fitter(shape, sigma_prior=sigma_prior)
theta, _, _, _ = fitter.fit(series.ages, series.single_expression)
thetas.append(theta)
fig = plot_one_series(series, shapes, thetas, yrange)
save_figure(fig,
'RP/fit-examples-{}-{}.png'.format(g, r),
under_results=True)
age_scaler = LogScaler()
pathway = '17full'
data = GeneData.load('both').restrict_pathway(pathway).scale_ages(age_scaler)
sigmoid = Sigslope(priors='sigslope80')
spline = Spline()
poly1 = Poly(1,priors='poly1')
poly2 = Poly(2,priors='poly2')
poly3 = Poly(3,priors='poly3')
shapes = [sigmoid, spline, poly1, poly2, poly3]
fitters = [Fitter(shape, sigma_prior='normal' if not shape.has_special_fitting() else None) for shape in shapes]
fits = [get_all_fits(data,fitter,allow_new_computation=False) for fitter in fitters]
fig = plot_comparison_bar(data, shapes, fits)
save_figure(fig,'RP/shape-comparison-bar-{}.png'.format(data.pathway), under_results=True)
#fig = plot_comparison_bar(data, shapes, fits, threshold_percentile=50)
#save_figure(fig,'RP/shape-comparison-bar-{}-top-half.png'.format(data.pathway), under_results=True)
fig = plot_comparison_over_R2_score(data, shapes, fits)
save_figure(fig,'RP/shape-comparison-vs-R2-{}.png'.format(data.pathway), under_results=True)
fig = plot_comparison_over_R2_score(data, shapes, fits, zoom=(0.3,1))
save_figure(fig,'RP/shape-comparison-vs-R2-{}-zoom.png'.format(data.pathway), under_results=True)
for i in xrange(1,len(shapes)):
fig = plot_comparison_scatter(data,shapes[0],fits[0],shapes[i],fits[i])
save_figure(fig,'RP/scatter-{}-{}-{}.png'.format(shapes[0],shapes[i],pathway), under_results=True)
plt.close('all')
def basic_fit():
print 'Drawing basic fit...'
theta,_,_,_ = fitter.fit(x,y)
fig = plot_one_series(series,shape,theta,yrange)
save_figure(fig,'RP/methods-1-basic-fit.png', under_results=True)
def annotate_parameters():
print 'Drawing fit with parameters...'
theta,_,_,_ = fitter.fit(x,y)
fig = plot_one_series(series,shape,theta, yrange, b_annotate=True)
save_figure(fig,'RP/methods-2-sigmoid-params.png', under_results=True)
def annotate_parameters():
print 'Drawing fit with parameters...'
theta, _, _, _ = fitter.fit(x, y)
fig = plot_one_series(series, shape, theta, yrange, b_annotate=True)
save_figure(fig, 'RP/methods-2-sigmoid-params.png', under_results=True)
disable_all_warnings()
cfg.verbosity = 1
age_scaler = LogScaler()
data = GeneData.load('both').restrict_pathway(pathway).scale_ages(age_scaler)
shape = Sigmoid(priors='sigmoid_wide')
fitter = Fitter(shape, sigma_prior='normal')
fits = get_all_fits(data, fitter, allow_new_computation=False)
dirname = 'bootstrap'
fits = add_change_distributions(data, fitter, fits, n_bins=n_bins)
fig = plot_bootstrap_onset_variance(data, fits)
save_figure(fig,
'{}/onset-variance-{}.png'.format(dirname, pathway),
under_results=True,
b_close=True)
fig = plot_change_width_scatter(data, fitter, fits)
save_figure(fig,
'{}/width-scatter-{}.png'.format(dirname, pathway),
under_results=True,
b_close=True)
for g, r in gene_regions:
ds_name = data.region_to_dataset()[r]
fit = fits[ds_name][(g, r)]
fig = plot_bootstrap_fits(data, fit, n_bins=n_bins, n_samples=n_samples)
save_figure(fig,
'{}/fits-{}-{}.png'.format(dirname, g, r),
under_results=True,
shape = Sigmoid('sigmoid_wide')
fitter = Fitter(shape, sigma_prior='normal')
fits = get_all_fits(data, fitter, allow_new_computation=False)
fits_shuffled = get_all_fits(data_shuffled,
fitter,
allow_new_computation=False)
R2_pairs = [(fit.LOO_score, fit2.LOO_score)
for fit, fit2 in iterate_fits(fits, fits_shuffled)]
R2 = np.array([r for r, r_shuffled in R2_pairs])
R2_shuffled = np.array([r_shuffled for r, r_shuffled in R2_pairs])
name = '{}-{}'.format(data.pathway, shape.cache_name())
fig = plot_score_distribution(R2, R2_shuffled)
save_figure(fig,
'RP/R2-distribution-{}.png'.format(name),
under_results=True,
b_close=True)
mu_shuffled = np.mean(R2_shuffled)
std_shuffled = np.std(R2_shuffled)
z_scores = (R2 - mu_shuffled) / std_shuffled
fig = plot_z_scores(z_scores)
save_figure(fig,
'RP/R2-z-scores-{}.png'.format(name),
under_results=True,
b_close=True)
T, signed_rank_p_value = wilcoxon(R2, R2_shuffled)
maxShuffled = R2_shuffled.max()
nAbove = np.count_nonzero(R2 > maxShuffled)
nTotal = len(R2)
age_scaler = LogScaler()
pathway = '17full'
data = GeneData.load('both').restrict_pathway(pathway).scale_ages(age_scaler)
data_shuffled = GeneData.load('both').restrict_pathway(pathway).scale_ages(age_scaler).shuffle()
shape = Sigmoid('sigmoid_wide')
fitter = Fitter(shape,sigma_prior='normal')
fits = get_all_fits(data,fitter,allow_new_computation=False)
fits_shuffled = get_all_fits(data_shuffled,fitter,allow_new_computation=False)
R2_pairs = [(fit.LOO_score,fit2.LOO_score) for fit,fit2 in iterate_fits(fits,fits_shuffled)]
R2 = np.array([r for r,r_shuffled in R2_pairs])
R2_shuffled = np.array([r_shuffled for r,r_shuffled in R2_pairs])
name = '{}-{}'.format(data.pathway,shape.cache_name())
fig = plot_score_distribution(R2,R2_shuffled)
save_figure(fig,'RP/R2-distribution-{}.png'.format(name), under_results=True, b_close=True)
mu_shuffled = np.mean(R2_shuffled)
std_shuffled = np.std(R2_shuffled)
z_scores = (R2-mu_shuffled)/std_shuffled
fig = plot_z_scores(z_scores)
save_figure(fig,'RP/R2-z-scores-{}.png'.format(name), under_results=True, b_close=True)
T, signed_rank_p_value = wilcoxon(R2, R2_shuffled)
maxShuffled = R2_shuffled.max()
nAbove = np.count_nonzero(R2 > maxShuffled)
nTotal = len(R2)
pct = 100.0 * nAbove/nTotal
filename = join(results_dir(),'RP/R2-distribution-{}.txt'.format(name))
with open(filename,'w') as f:
print('shuffled = {:.2g} +/- {:.2g}'.format(mu_shuffled,std_shuffled), file=f)
for pathway in lst_pathways:
data = GeneData.load('both').restrict_pathway(pathway).scale_ages(
age_scaler)
shape = Sigmoid(priors='sigmoid_wide')
fitter = Fitter(shape, sigma_prior='normal')
fits = get_all_fits(data, fitter, allow_new_computation=False)
# R2_threshold = 0.5 YYY problem - we might be using bad fits.
regions = ['OFC', 'M1C', 'S1C', 'IPC', 'V1C']
scores = []
for g in data.gene_names:
r, pval, lst_R2 = get_gene_correlation(fits, g, regions)
scores.append((g, r, pval, lst_R2))
fig = plot_correlation_histogram(scores, pathway)
save_figure(fig,
'{}/gradual-maturation-hist.png'.format(pathway, pathway),
under_results=True,
b_close=True)
for fR2 in [np.mean]: #[min,max,np.mean]:
fig = plot_scatter(scores, pathway, fR2)
save_figure(fig,
'{}/gradual-maturation-scatter-{}.png'.format(
pathway, fR2.__name__),
under_results=True,
b_close=True)
create_top_correlations_html(data, fitter, fits, scores, regions)
nScores, yScores = zip(*score_pairs)
_, pval = scipy.stats.wilcoxon(nScores, yScores)
pval = pval/2 # one sided p-value
print '*** wilcoxon signed rank p-value (one sided) = {:.3g}'.format(pval)
# find examples of best improvements
diffs = [(f2.LOO_score-f1.LOO_score, f1.LOO_score, f2.LOO_score, g, r) for dsname,g,r,f1,f2 in iterate_fits(nFits, yFits, R2_threshold=-1, return_keys=True)]
diffs.sort(reverse=True)
print 'Gene/Regions for which priors produce best R2 improvement:'
for i,(delta,R2_without, R2_with, g,r) in enumerate(diffs[:10]):
print '{i}) {g}@{r}, delta-R2={delta:.3g}. R2_without={R2_without:.3g}, R2_with={R2_with:.3g}'.format(**locals())
cfg.verbosity = 1
age_scaler = LogScaler()
pathway = '17full'
data = GeneData.load('both').restrict_pathway(pathway).scale_ages(age_scaler)
priors_name = 'sigslope80'
analyze_paired_scores_with_and_without_priors()
variations = [analyze_variant(t,s) for t,s in product([False,True],[False,True])]
fig = plot_bar(variations)
save_figure(fig,'RP/prior-variations-bar.png', under_results=True)
fig = plot_pctiles(variations, min_q=5)
save_figure(fig,'RP/prior-variations-percentiles.png', under_results=True)
#fig = plot_theta_diff_scatter()
#save_figure(fig,'RP/prior-variations-scatter.png', under_results=True)
shape = Sigmoid(priors='sigmoid_wide')
fitter = Fitter(shape, sigma_prior='normal')
fits = get_all_fits(data, fitter)
R2_threshold = 0.5
for b_unique in [False, True]:
dct_pathways = load_17_pathways_breakdown(b_unique)
dct_pathways['17 pathways'] = None
for name, genes in dct_pathways.iteritems():
fig = plot_onset_times(all_data, data, fitter, fits, {name: genes},
R2_threshold, b_unique)
str_dir = 'unique' if b_unique else 'overlapping'
str_unique = ' (unique)' if b_unique else ''
filename = 'RP/{}/change-distributions-{}{}.png'.format(
str_dir, name, str_unique)
save_figure(fig, filename, under_results=True)
# selected plots
lst_pathways = [
'17 pathways', 'Amphetamine addiction', 'Cholinergic synapse',
'Cocaine addiction', 'Glutamatergic synapse'
]
dct_pathways = {k: dct_pathways[k] for k in lst_pathways}
fig = plot_onset_times(all_data, data, fitter, fits, dct_pathways,
R2_threshold, b_unique)
str_dir = 'unique' if b_unique else 'overlapping'
str_unique = ' (unique)' if b_unique else ''
filename = 'RP/{}/selected-change-distributions{}.png'.format(
str_dir, str_unique)
save_figure(fig, filename, under_results=True)
shapes = [sigmoid, sigslope]
fitters = [Fitter(shape, sigma_prior='normal') for shape in shapes]
#####################################################
# Example fits
#####################################################
GRs = [
('ABHD4','STC', (5, 8)),
]
for g,r,yrange in GRs:
for fitter in fitters:
print 'Doing {}@{}...'.format(g,r)
series = data.get_one_series(g,r)
theta,_,_,_ = fitter.fit(series.ages, series.single_expression)
fig = plot_one_series(series, fitter.shape, theta, yrange)
save_figure(fig,'RP/fit-examples-{}-{}-{}.png'.format(fitter.shape.cache_name(), g,r), under_results=True)
#####################################################
# Comparison for whole pathway
#####################################################
pathway = '17full'
data = data.restrict_pathway(pathway)
fits = [get_all_fits(data,fitter,allow_new_computation=False) for fitter in fitters]
fig = plot_comparison_bar(data, shapes, fits)
save_figure(fig,'RP/sigslope-comparison-bar-{}.png'.format(data.pathway), under_results=True)
fig = plot_comparison_over_R2_score(data, shapes, fits)
save_figure(fig,'RP/sigslope-comparison-vs-R2-{}.png'.format(data.pathway), under_results=True)
if __name__ == '__main__':
disable_all_warnings()
parser = get_common_parser()
parser.add_argument('--shape2', required=True, help='The shape to compare against', choices=allowed_shape_names())
parser.add_argument('--scaling2', help='The scaling used when fitting shape2. Default: none', choices=allowed_scaler_names())
parser.add_argument('--sigma_prior2', help='Prior to use for 1/sigma when fitting shape2. Default: None', choices=get_allowed_priors(is_sigma=True))
parser.add_argument('--priors2', help='The priors used for theta when fitting shape2. Default: None', choices=get_allowed_priors())
parser.add_argument('--filename', help='Where to save the figure. Default: results/comparison.png')
parser.add_argument('--show', help='Show figure and wait before exiting', action='store_true')
parser.add_argument('--ndiffs', type=int, default=5, help='Number of top diffs to show. Default=5.')
args = parser.parse_args()
data1, fitter1 = process_common_inputs(args)
data2 = get_data_from_args(args.dataset, args.pathway, args.from_age, args.scaling2, args.shuffle)
fitter2 = get_fitter_from_args(args.shape2, args.priors2, args.sigma_prior2)
fits1 = get_all_fits(data1,fitter1)
fits2 = get_all_fits(data2,fitter2)
print_diff_points(data1,fitter1,fits1, data2,fitter2,fits2, args.ndiffs)
fig = plot_comparison_scatter(data1,fitter1,fits1, data2,fitter2,fits2)
filename = args.filename
if filename is None:
ensure_dir(results_dir())
filename = join(results_dir(), 'shape_comparison.png')
save_figure(fig, filename)
if args.show:
plt.show(block=True)
def show_loo_score():
print 'Drawing LOO prediction for all points and R2 score...'
theta,_,test_preds,_ = fitter.fit(x,y,loo=True)
fig = plot_one_series(series,shape,theta=None,yrange=yrange,test_preds=test_preds)
save_figure(fig,'RP/methods-4-R2-score.png', under_results=True)
ticks = ax.get_yticks()
ticks = np.array([ticks[0], ticks[-1]])
ax.set_yticks(ticks)
ax.set_yticklabels(['{:g}'.format(t) for t in ticks], fontsize=fontsize)
return fig
cfg.verbosity = 1
age_scaler = LogScaler()
data = GeneData.load('both').scale_ages(age_scaler)
shapes = [Sigmoid('sigmoid_wide'), Poly(1,'poly1'), Poly(3,'poly3'), Spline()]
GRs = [
('ADRB1','A1C', (5, 8)),
('GLRA2','STC', (5, 12)),
('TUBA1A','V1C', (10, 14)),
]
for g,r,yrange in GRs:
print 'Doing {}@{}...'.format(g,r)
thetas = []
for shape in shapes:
series = data.get_one_series(g,r)
sigma_prior = 'normal' if not isinstance(shape,Spline) else None
fitter = Fitter(shape, sigma_prior=sigma_prior)
theta,_,_,_ = fitter.fit(series.ages, series.single_expression)
thetas.append(theta)
fig = plot_one_series(series,shapes,thetas,yrange)
save_figure(fig,'RP/fit-examples-{}-{}.png'.format(g,r), under_results=True)
print '*** wilcoxon signed rank p-value (one sided) = {:.3g}'.format(pval)
# find examples of best improvements
diffs = [(f2.LOO_score - f1.LOO_score, f1.LOO_score, f2.LOO_score, g, r)
for dsname, g, r, f1, f2 in iterate_fits(
nFits, yFits, R2_threshold=-1, return_keys=True)]
diffs.sort(reverse=True)
print 'Gene/Regions for which priors produce best R2 improvement:'
for i, (delta, R2_without, R2_with, g, r) in enumerate(diffs[:10]):
print '{i}) {g}@{r}, delta-R2={delta:.3g}. R2_without={R2_without:.3g}, R2_with={R2_with:.3g}'.format(
**locals())
cfg.verbosity = 1
age_scaler = LogScaler()
pathway = '17full'
data = GeneData.load('both').restrict_pathway(pathway).scale_ages(age_scaler)
priors_name = 'sigslope80'
analyze_paired_scores_with_and_without_priors()
variations = [
analyze_variant(t, s) for t, s in product([False, True], [False, True])
]
fig = plot_bar(variations)
save_figure(fig, 'RP/prior-variations-bar.png', under_results=True)
fig = plot_pctiles(variations, min_q=5)
save_figure(fig, 'RP/prior-variations-percentiles.png', under_results=True)
#fig = plot_theta_diff_scatter()
#save_figure(fig,'RP/prior-variations-scatter.png', under_results=True)
