def plot_acc_quartiles(acc_df, args, cdata):
mpl.rcParams['axes.linewidth'] = 1.2
mpl.rcParams['axes.edgecolor'] = '0.05'
fig, (ax_auc, ax_aupr) = plt.subplots(figsize=(22, 10), ncols=2)
mtype_sizes = [len(cdata.train_mut[gene]) / len(cdata.samples)
for gene in acc_df.index]
auc_vals = acc_df['AUC'].quantile(q=0.25, axis=1)
aupr_vals = acc_df['AUPR'].quantile(q=0.25, axis=1)
ax_auc.scatter(mtype_sizes, auc_vals, s=15, c='black', alpha=0.47)
ax_aupr.scatter(mtype_sizes, aupr_vals, s=15, c='black', alpha=0.47)
auc_annot = place_annot(mtype_sizes, auc_vals.values.tolist(),
size_vec=[15 for _ in mtype_sizes],
annot_vec=aupr_vals.index, x_range=1, y_range=1)
for annot_x, annot_y, annot, halign in auc_annot:
ax_auc.text(annot_x, annot_y, annot, size=11, ha=halign)
aupr_annot = place_annot(mtype_sizes, aupr_vals.values.tolist(),
size_vec=[15 for _ in mtype_sizes],
annot_vec=aupr_vals.index, x_range=1, y_range=1)
for annot_x, annot_y, annot, halign in aupr_annot:
ax_aupr.text(annot_x, annot_y, annot, size=11, ha=halign)
for ax in (ax_auc, ax_aupr):
ax.set_xlim(0, 1)
ax.set_ylim(0, 1)
ax_auc.plot([-1, 2], [0.5, 0.5],
linewidth=1.7, linestyle='--', color='#550000', alpha=0.6)
ax_aupr.plot([-1, 2], [-1, 2],
linewidth=1.7, linestyle='--', color='#550000', alpha=0.6)
fig.text(0.5, -0.03,
'Proportion of {} Samples Mutated'.format(args.cohort),
ha='center', va='center', fontsize=22, weight='semibold')
ax_auc.set_ylabel('1st Quartile AUC', fontsize=22, weight='semibold')
ax_aupr.set_ylabel('1st Quartile AUPR', fontsize=22, weight='semibold')
fig.tight_layout(w_pad=2.2, h_pad=5.1)
fig.savefig(
os.path.join(plot_dir, args.model_name.split('__')[0],
'{}__acc-quartiles__{}-{}_samps-{}.png'.format(
args.model_name.split('__')[1], args.expr_source,
args.cohort, args.samp_cutoff
)),
dpi=250, bbox_inches='tight'
)
plt.close()
def plot_auc_quartiles(auc_df, args):
mpl.rcParams['axes.linewidth'] = 1.2
mpl.rcParams['axes.edgecolor'] = '0.05'
fig, ax = plt.subplots(figsize=(14, 13))
test_aucs = auc_df.applymap(itemgetter('test'))
quart_df = pd.DataFrame(index=auc_df.index, columns=['Min', 'Max'])
new_indx = ['' for _ in auc_df.index]
for i, (((coh1, coh2), mtype),
auc_dicts) in enumerate(test_aucs.iterrows()):
auc_quants = pd.DataFrame.from_records(
auc_dicts.values).quantile(q=0.25).sort_values()
new_indx[i] = "({}) {}".format(' x '.join(auc_quants.index),
str(mtype))
quart_df.iloc[i, :] = auc_quants.values
quart_df.index = new_indx
plot_min = np.min(quart_df.values) - 0.01
ax.scatter(quart_df.Min, quart_df.Max, s=15, c='black', alpha=0.47)
for annot_x, annot_y, annot, halign in place_annot(
quart_df.Min.tolist(),
quart_df.Max.tolist(),
size_vec=[15 for _ in auc_df.index],
annot_vec=quart_df.index,
x_range=1 - plot_min,
y_range=1 - plot_min,
gap_adj=79):
ax.text(annot_x, annot_y, annot, size=11, ha=halign)
ax.tick_params(pad=5.1)
ax.set_xlim(plot_min, 1)
ax.set_ylim(plot_min, 1)
ax.set_xlabel('1st Qrt. AUC, min cohort', fontsize=22, weight='semibold')
ax.set_ylabel('1st Qrt. AUC, max cohort', fontsize=22, weight='semibold')
ax.plot([-1, 2], [-1, 2],
linewidth=1.7,
linestyle='--',
color='#550000',
alpha=0.6)
fig.savefig(os.path.join(
plot_dir,
args.model_name.split('__')[0],
'{}__acc-quartiles__{}_samps-{}.png'.format(
args.model_name.split('__')[1], args.expr_source,
args.samp_cutoff)),
dpi=250,
bbox_inches='tight')
plt.close()
def plot_tuning_gene(par_df, acc_df, use_clf, args, cdata):
fig, axarr = plt.subplots(figsize=(13, 12 * len(use_clf.tune_priors)),
nrows=len(use_clf.tune_priors), ncols=1,
squeeze=False)
for ax, (par_name, tune_distr) in zip(axarr.flatten(),
use_clf.tune_priors):
par_vals = par_df[par_name].groupby(level=0).median()
acc_vals = acc_df['AUC'].quantile(q=0.25, axis=1)
size_vec = [1073 * len(cdata.train_mut[gene]) / len(cdata.samples)
for gene in acc_vals.index]
if detect_log_distr(tune_distr):
par_vals = np.log10(par_vals)
plt_xmin = 2 * np.log10(tune_distr[0]) - np.log10(tune_distr[1])
plt_xmax = 2 * np.log10(tune_distr[-1]) - np.log10(tune_distr[-2])
else:
plt_xmin = 2 * tune_distr[0] - tune_distr[1]
plt_xmax = 2 * tune_distr[-1] - tune_distr[-2]
par_vals += np.random.normal(
0, (plt_xmax - plt_xmin) / (len(tune_distr) * 19), acc_df.shape[0])
ax.scatter(par_vals, acc_vals, s=size_vec, c='black', alpha=0.23)
ax.set_xlim(plt_xmin, plt_xmax)
ax.set_ylim(0, 1)
ax.axhline(y=0.5, color='#550000',
linewidth=3.1, linestyle='--', alpha=0.32)
annot_placed = place_annot(
par_vals, acc_vals.values.tolist(), size_vec=size_vec,
annot_vec=acc_vals.index, x_range=plt_xmax - plt_xmin, y_range=1
)
for annot_x, annot_y, annot, halign in annot_placed:
ax.text(annot_x, annot_y, annot, size=11, ha=halign)
ax.set_xlabel('Median Tuned {} Value'.format(par_name),
fontsize=26, weight='semibold')
ax.set_ylabel('1st Quartile AUC', fontsize=26, weight='semibold')
plt.tight_layout()
fig.savefig(
os.path.join(plot_dir, args.model_name.split('__')[0],
'{}__tuning-gene__{}-{}_samps-{}.png'.format(
args.model_name.split('__')[1], args.expr_source,
args.cohort, args.samp_cutoff
)),
dpi=250, bbox_inches='tight'
)
plt.close()
def plot_aupr_time(out_dict, args):
fig, axarr = plt.subplots(figsize=(9, 15), nrows=2, sharex=True)
time_quarts = np.log2(
pd.Series({
mdl: (out_data['Tune']['Time']['fit']['avg'] +
out_data['Tune']['Time']['fit']['std']).groupby(
axis=1, level=0).quantile(q=0.75).mean().mean()
for mdl, out_data in out_dict.items()
}))
aupr_vals = {
mdl: out_data['Fit']['test']['AUPR'].quantile(q=0.25, axis=1)
for mdl, out_data in out_dict.items()
}
aupr_list = [
pd.Series({mdl: vals.mean()
for mdl, vals in aupr_vals.items()}),
pd.Series(
{mdl: vals.quantile(q=0.75)
for mdl, vals in aupr_vals.items()}),
]
expr_vec = time_quarts.index.get_level_values(0)
expr_shapes = [
use_marks[sorted(set(expr_vec)).index(expr)] for expr in expr_vec
]
model_vec = time_quarts.index.get_level_values(1).str.split('__').map(
itemgetter(0))
model_cmap = sns.color_palette('Set1',
n_colors=len(set(model_vec)),
desat=.34)
model_clrs = [
model_cmap[sorted(set(model_vec)).index(mdl)] for mdl in model_vec
]
for ax, auprs in zip(axarr, aupr_list):
for time_val, aupr_val, expr_shape, model_clr in zip(
time_quarts.values, auprs.values, expr_shapes, model_clrs):
ax.scatter(time_val,
aupr_val,
marker=expr_shape,
c=model_clr,
s=71,
alpha=0.41)
for annot_x, annot_y, annot, halign in place_annot(
time_quarts.values.tolist(),
auprs.values.tolist(),
size_vec=[71 for _ in time_quarts],
annot_vec=[' '.join(tst) for tst in time_quarts.index],
x_range=time_quarts.max() - time_quarts.min(),
y_range=auprs.max() - auprs.min(),
gap_adj=79):
ax.text(annot_x, annot_y, annot, size=10, ha=halign)
ax.tick_params(axis='y', labelsize=14)
axarr[1].xaxis.set_major_formatter(ticker.FormatStrFormatter(r'$2^{%d}$'))
axarr[1].tick_params(axis='x', labelsize=21, pad=7)
axarr[0].set_ylabel('Average AUPR', size=23, weight='semibold')
axarr[1].set_ylabel('Third Quartile AUPR', size=23, weight='semibold')
plt.xlabel('Fitting Time (seconds)', size=23, weight='semibold')
plt.tight_layout(h_pad=3.3)
fig.savefig(os.path.join(plot_dir,
'{}__aupr-time.svg'.format(args.cohort)),
bbox_inches='tight',
format='svg')
plt.close()
def plot_tuning_mtype_grid(par_df, auc_df, use_clf, args, cdata):
par_count = len(use_clf.tune_priors)
fig, axarr = plt.subplots(figsize=(0.5 + 7 * par_count, 7 * par_count),
nrows=par_count,
ncols=par_count)
auc_vals = auc_df.quantile(q=0.25, axis=1)
auc_clrs = auc_vals.apply(auc_cmap)
size_vec = [
461 * sum(cdata.train_pheno(mtype)) /
(len(cdata.get_samples()) * par_count) for mtype in auc_vals.index
]
for i, (par_name, tune_distr) in enumerate(use_clf.tune_priors):
axarr[i, i].grid(False)
if detect_log_distr(tune_distr):
use_distr = [np.log10(par_val) for par_val in tune_distr]
par_lbl = par_name + '\n(log-scale)'
else:
use_distr = tune_distr
par_lbl = par_name
distr_diff = np.mean(
np.array(use_distr[1:]) - np.array(use_distr[:-1]))
plt_min = use_distr[0] - distr_diff / 2
plt_max = use_distr[-1] + distr_diff / 2
axarr[i, i].set_xlim(plt_min, plt_max)
axarr[i, i].set_ylim(plt_min, plt_max)
axarr[i, i].text((plt_min + plt_max) / 2, (plt_min + plt_max) / 2,
par_lbl,
ha='center',
fontsize=28,
weight='semibold')
for par_val in use_distr:
axarr[i, i].axhline(y=par_val,
color='#116611',
ls='--',
linewidth=4.1,
alpha=0.27)
axarr[i, i].axvline(x=par_val,
color='#116611',
ls='--',
linewidth=4.1,
alpha=0.27)
for (i, (par_name1, tn_distr1)), (j, (par_name2, tn_distr2)) in combn(
enumerate(use_clf.tune_priors), 2):
if detect_log_distr(tn_distr1):
use_distr1 = [np.log10(par_val) for par_val in tn_distr1]
par_meds1 = np.log10(par_df[par_name1]).median(axis=1)
par_means1 = np.log10(par_df[par_name1]).mean(axis=1)
distr_diff = np.mean(
np.log10(np.array(tn_distr1[1:])) -
np.log10(np.array(tn_distr1[:-1])))
plt_ymin = np.log10(tn_distr1[0]) - distr_diff / 2
plt_ymax = np.log10(tn_distr1[-1]) + distr_diff / 2
else:
use_distr1 = tn_distr1
par_meds1 = par_df[par_name1].median(axis=1)
par_means1 = par_df[par_name1].mean(axis=1)
distr_diff = np.mean(
np.array(tn_distr1[1:]) - np.array(tn_distr1[:-1]))
plt_ymin = tn_distr1[0] - distr_diff / 2
plt_ymax = tn_distr1[-1] + distr_diff / 2
if detect_log_distr(tn_distr2):
use_distr2 = [np.log10(par_val) for par_val in tn_distr2]
par_meds2 = np.log10(par_df[par_name2]).median(axis=1)
par_means2 = np.log10(par_df[par_name2]).mean(axis=1)
distr_diff = np.mean(
np.log10(np.array(tn_distr2[1:])) -
np.log10(np.array(tn_distr2[:-1])))
plt_xmin = np.log10(tn_distr2[0]) - distr_diff / 2
plt_xmax = np.log10(tn_distr2[-1]) + distr_diff / 2
else:
use_distr2 = tn_distr2
par_meds2 = par_df[par_name2].median(axis=1)
par_means2 = par_df[par_name2].mean(axis=1)
distr_diff = np.mean(
np.array(tn_distr2[1:]) - np.array(tn_distr2[:-1]))
plt_xmin = tn_distr2[0] - distr_diff / 2
plt_xmax = tn_distr2[-1] + distr_diff / 2
par_meds1 = par_meds1[auc_clrs.index]
par_meds2 = par_meds2[auc_clrs.index]
y_adj = (plt_ymax - plt_ymin) / len(tn_distr1)
x_adj = (plt_xmax - plt_xmin) / len(tn_distr2)
plt_adj = (plt_xmax - plt_xmin) / (plt_ymax - plt_ymin)
for med1, med2 in set(zip(par_meds1, par_meds2)):
use_indx = (par_meds1 == med1) & (par_meds2 == med2)
cnt_adj = use_indx.sum()**0.49
use_sizes = [s for s, ix in zip(size_vec, use_indx) if ix]
sort_indx = sorted(enumerate(use_sizes),
key=lambda x: x[1],
reverse=True)
from circlify import circlify
mpl.use('Agg')
for k, circ in enumerate(circlify([s for _, s in sort_indx])):
axarr[i, j].scatter(
med2 + (1 / 23) * cnt_adj * circ.y * plt_adj,
med1 + (1 / 23) * cnt_adj * circ.x * plt_adj**-1,
s=sort_indx[k][1],
c=auc_clrs[use_indx][sort_indx[k][0]],
alpha=0.36,
edgecolor='black')
par_means1 += np.random.normal(0, y_adj / 27, auc_df.shape[0])
par_means2 += np.random.normal(0, x_adj / 27, auc_df.shape[0])
axarr[j, i].scatter(par_means1[auc_clrs.index],
par_means2[auc_clrs.index],
s=size_vec,
c=auc_clrs,
alpha=0.36,
edgecolor='black')
axarr[i, j].set_xlim(plt_xmin, plt_xmax)
axarr[i, j].set_ylim(plt_ymin, plt_ymax)
axarr[j, i].set_ylim(plt_xmin, plt_xmax)
axarr[j, i].set_xlim(plt_ymin, plt_ymax)
annot_placed = place_annot(par_meds2,
par_meds1,
size_vec=size_vec,
annot_vec=auc_vals.index,
x_range=plt_xmax - plt_xmin,
y_range=plt_ymax - plt_ymin)
for annot_x, annot_y, annot, halign in annot_placed:
axarr[i, j].text(annot_x, annot_y, annot, size=11, ha=halign)
for par_val1 in use_distr1:
axarr[i, j].axhline(y=par_val1,
color='#116611',
ls=':',
linewidth=2.3,
alpha=0.19)
axarr[j, i].axvline(x=par_val1,
color='#116611',
ls=':',
linewidth=2.3,
alpha=0.19)
for par_val2 in use_distr2:
axarr[i, j].axvline(x=par_val2,
color='#116611',
ls=':',
linewidth=2.3,
alpha=0.19)
axarr[j, i].axhline(y=par_val2,
color='#116611',
ls=':',
linewidth=2.3,
alpha=0.19)
plt.tight_layout()
fig.savefig(os.path.join(
plot_dir, args.expr_source,
"{}__samps-{}".format(args.cohort, args.samp_cutoff),
args.model_name.split('__')[0],
"{}__tuning-mtype-grid.svg".format(args.model_name.split('__')[1])),
bbox_inches='tight',
format='svg')
plt.close()
def plot_tuning_mtype(par_df, auc_df, use_clf, args, cdata):
fig, axarr = plt.subplots(figsize=(1 + 9 * len(use_clf.tune_priors), 13),
nrows=3,
ncols=len(use_clf.tune_priors),
gridspec_kw={'height_ratios': [1, 0.3, 1]},
squeeze=False,
sharex=False,
sharey=True)
auc_vals = auc_df.quantile(q=0.25, axis=1)
size_vec = [
198 * len(mtype.get_samples(cdata.mtree)) / len(cdata.get_samples())
for mtype in auc_vals.index
]
for i, (par_name, tune_distr) in enumerate(use_clf.tune_priors):
axarr[1, i].set_axis_off()
axarr[2, i].tick_params(length=6)
if detect_log_distr(tune_distr):
med_vals = np.log10(par_df[par_name]).median(axis=1)
mean_vals = np.log10(par_df[par_name]).mean(axis=1)
use_distr = [np.log10(par_val) for par_val in tune_distr]
par_lbl = par_name + '\n(log-scale)'
else:
med_vals = par_df[par_name].median(axis=1)
mean_vals = par_df[par_name].mean(axis=1)
use_distr = tune_distr
par_lbl = par_name
med_vals = med_vals[auc_vals.index]
mean_vals = mean_vals[auc_vals.index]
distr_diff = np.mean(
np.array(use_distr[1:]) - np.array(use_distr[:-1]))
for j in range(3):
axarr[j, i].set_xlim(use_distr[0] - distr_diff / 2,
use_distr[-1] + distr_diff / 2)
axarr[1, i].text((use_distr[0] + use_distr[-1]) / 2,
0.5,
par_lbl,
ha='center',
va='center',
fontsize=25,
weight='semibold')
med_vals += np.random.normal(0, (use_distr[-1] - use_distr[0]) /
(len(tune_distr) * 17), auc_df.shape[0])
mean_vals += np.random.normal(0, (use_distr[-1] - use_distr[0]) /
(len(tune_distr) * 23), auc_df.shape[0])
axarr[0, i].scatter(med_vals,
auc_vals,
s=size_vec,
c='black',
alpha=0.23)
axarr[2, i].scatter(mean_vals,
auc_vals,
s=size_vec,
c='black',
alpha=0.23)
axarr[0, i].set_ylim(0, 1)
axarr[2, i].set_ylim(0, 1)
axarr[0, i].set_ylabel("1st Quartile AUC", size=19, weight='semibold')
axarr[2, i].set_ylabel("1st Quartile AUC", size=19, weight='semibold')
axarr[0, i].axhline(y=0.5,
color='#550000',
linewidth=2.3,
linestyle='--',
alpha=0.32)
axarr[2, i].axhline(y=0.5,
color='#550000',
linewidth=2.3,
linestyle='--',
alpha=0.32)
for par_val in use_distr:
axarr[1, i].axvline(x=par_val,
color='#116611',
ls='--',
linewidth=3.4,
alpha=0.27)
axarr[0, i].axvline(x=par_val,
color='#116611',
ls=':',
linewidth=1.3,
alpha=0.16)
axarr[2, i].axvline(x=par_val,
color='#116611',
ls=':',
linewidth=1.3,
alpha=0.16)
annot_placed = place_annot(med_vals,
auc_vals.values.tolist(),
size_vec=size_vec,
annot_vec=auc_vals.index,
x_range=use_distr[-1] - use_distr[0] +
2 * distr_diff,
y_range=1)
for annot_x, annot_y, annot, halign in annot_placed:
axarr[0, i].text(annot_x, annot_y, annot, size=8, ha=halign)
plt.tight_layout(h_pad=0)
fig.savefig(os.path.join(
plot_dir, args.expr_source,
"{}__samps-{}".format(args.cohort, args.samp_cutoff),
args.model_name.split('__')[0],
"{}__tuning-mtype.svg".format(args.model_name.split('__')[1])),
bbox_inches='tight',
format='svg')
plt.close()
def plot_acc_quartiles(auc_df, aupr_df, args, cdata):
mpl.rcParams['axes.linewidth'] = 1.2
mpl.rcParams['axes.edgecolor'] = '0.05'
fig, (ax_auc, ax_aupr) = plt.subplots(figsize=(22, 10), ncols=2)
auc_vals = auc_df.quantile(q=0.25, axis=1)
aupr_vals = aupr_df.quantile(q=0.25, axis=1)
mtype_sizes = [
len(mtype.get_samples(cdata.mtree)) / len(cdata.get_samples())
for mtype in auc_df.index
]
ax_auc.scatter(mtype_sizes, auc_vals, s=17, c='black', alpha=0.47)
ax_aupr.scatter(mtype_sizes, aupr_vals, s=17, c='black', alpha=0.47)
for annot_x, annot_y, annot, halign in place_annot(
mtype_sizes,
auc_vals.values.tolist(),
size_vec=[15 for _ in mtype_sizes],
annot_vec=aupr_vals.index,
x_range=max(mtype_sizes) * 1.03,
y_range=1,
gap_adj=53):
ax_auc.text(annot_x, annot_y, annot, size=11, ha=halign)
for annot_x, annot_y, annot, halign in place_annot(
mtype_sizes,
aupr_vals.values.tolist(),
size_vec=[15 for _ in mtype_sizes],
annot_vec=aupr_vals.index,
x_range=1,
y_range=1,
gap_adj=53):
ax_aupr.text(annot_x, annot_y, annot, size=11, ha=halign)
ax_auc.set_xlim(0, max(mtype_sizes) * 1.03)
ax_aupr.set_xlim(0, 1)
for ax in (ax_auc, ax_aupr):
ax.tick_params(pad=3.9)
ax.set_ylim(0, 1)
ax_auc.plot([-1, 2], [0.5, 0.5],
linewidth=1.7,
linestyle='--',
color='#550000',
alpha=0.6)
ax_aupr.plot([-1, 2], [-1, 2],
linewidth=1.7,
linestyle='--',
color='#550000',
alpha=0.6)
fig.text(0.5,
-0.03,
'Proportion of {} Samples Mutated'.format(args.cohort),
ha='center',
va='center',
fontsize=22,
weight='semibold')
ax_auc.set_ylabel('1st Quartile AUC', fontsize=22, weight='semibold')
ax_aupr.set_ylabel('1st Quartile AUPR', fontsize=22, weight='semibold')
fig.tight_layout(w_pad=2.2, h_pad=5.1)
fig.savefig(os.path.join(
plot_dir, args.expr_source,
"{}__samps-{}".format(args.cohort, args.samp_cutoff),
args.model_name.split('__')[0],
"{}__acc-quartiles.svg".format(args.model_name.split('__')[1])),
bbox_inches='tight',
format='svg')
plt.close()
