-
Notifications
You must be signed in to change notification settings - Fork 9
/
generate_bar_plots_2.py
217 lines (197 loc) · 9.53 KB
/
generate_bar_plots_2.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
import os
import re
import sys
import warnings
from argparse import ArgumentParser
warnings.filterwarnings('ignore', category=FutureWarning,
module='rpy2.robjects.pandas2ri')
import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
import seaborn as sns
from pandas.api.types import is_object_dtype, is_string_dtype
import rpy2.rinterface_lib.embedded as r_embedded
r_embedded.set_initoptions(
('rpy2', '--quiet', '--no-save', '--max-ppsize=500000'))
import rpy2.robjects as robjects
from joblib import dump, load
from matplotlib import ticker
from matplotlib.container import BarContainer
from rpy2.robjects import numpy2ri, pandas2ri
from rpy2.robjects.packages import importr
numpy2ri.activate()
pandas2ri.activate()
# suppress linux conda qt5 wayland warning
if sys.platform.startswith('linux'):
os.environ['XDG_SESSION_TYPE'] = 'x11'
parser = ArgumentParser()
parser.add_argument('--results-dir', type=str, default='results',
help='results dir')
parser.add_argument('--out-dir', type=str, default='figures/bar',
help='out dir')
parser.add_argument('--filter', type=str, choices=['signif', 'all'],
default='signif',
help='response model filter')
parser.add_argument('--file-format', type=str, nargs='+',
choices=['png', 'pdf', 'svg', 'tif'], default=['pdf'],
help='save file format')
args = parser.parse_args()
model_results_dir = '{}/models'.format(args.results_dir)
analysis_results_dir = '{}/analysis'.format(args.results_dir)
os.makedirs(args.out_dir, mode=0o755, exist_ok=True)
# for stripplot jitter
random_seed = 777
np.random.seed(random_seed)
data_types = ['kraken', 'htseq', 'combo']
metrics = ['roc_auc', 'pr_auc', 'balanced_accuracy']
metric_label = {'roc_auc': 'AUROC',
'pr_auc': 'AUPRC',
'balanced_accuracy': 'BCR'}
model_codes = ['rfe', 'lgr', 'edger', 'limma']
bcolors = ['#009E73', '#F0E442', '#0072B2']
ecolor = 'darkorange'
bcolors = sns.color_palette(bcolors)
title_fontsize = 16
x_axis_fontsize = 5 if args.filter == 'all' else 10
y_axis_fontsize = 12
label_fontsize = 12
legend_fontsize = 12
fig_height = 4
fig_width = 10 if args.filter == 'all' else 5
fig_dpi = 300
x_label_rotation = 60 if args.filter == 'all' else 45
plt.rcParams['figure.max_open_warning'] = 0
plt.rcParams['font.family'] = 'sans-serif'
plt.rcParams['font.sans-serif'] = ['Helvetica', 'Nimbus Sans', 'Arial',
'DejaVu Sans', 'sans-serif']
r_base = importr('base')
all_stats = pd.read_csv(
'{}/compared_runs.txt'.format(analysis_results_dir), sep='\t')
all_stats = all_stats.apply(
lambda x: x.str.lower() if is_object_dtype(x) or is_string_dtype(x) else x)
all_stats = all_stats.loc[all_stats['analysis'] == 'resp']
all_stats = all_stats.sort_values(by=['cancer', 'versus', 'features', 'how'])
signif_hits = pd.read_csv(
'{}/goodness_hits.txt'.format(analysis_results_dir), sep='\t')
signif_hits = signif_hits.apply(
lambda x: x.str.lower() if is_object_dtype(x) or is_string_dtype(x) else x)
signif_hits = signif_hits.loc[signif_hits['analysis'] == 'resp']
signif_hits = signif_hits.sort_values(
by=['cancer', 'versus', 'features', 'how'])
signif_hits = signif_hits.loc[signif_hits.duplicated(
subset=['cancer', 'versus', 'features'], keep=False)]
score_dfs, p_adjs = {}, {}
model_codes_regex = '|'.join(model_codes)
split_results_regex = re.compile(
'^(.+?_(?:{}))_split_results\\.pkl$'.format(model_codes_regex))
for dirpath, dirnames, filenames in sorted(os.walk(model_results_dir)):
for filename in filenames:
if m := re.search(split_results_regex, filename):
model_name = m.group(1)
_, cancer, analysis, target, data_type, *rest = (
model_name.split('_'))
if data_type == 'htseq':
model_code = '_'.join(rest[1:])
else:
model_code = '_'.join(rest)
if (args.filter == 'signif'
and not ((signif_hits['cancer'] == cancer)
& (signif_hits['versus'] == target)
& (signif_hits['features'] == data_type)).any()):
continue
if data_type not in p_adjs:
p_adjs[data_type] = {}
if model_code not in p_adjs[data_type]:
p_adjs[data_type][model_code] = []
p_adj = all_stats.loc[(all_stats['cancer'] == cancer)
& (all_stats['versus'] == target)
& (all_stats['features'] == data_type)
& (all_stats['how'] == model_code),
'p_adj'].item()
p_adjs[data_type][model_code].append(p_adj)
split_results_file = '{}/{}'.format(dirpath, filename)
print('Loading', split_results_file)
split_results = load(split_results_file)
for metric in metrics:
scores = []
for split_result in split_results:
if split_result is None:
scores.append(np.nan)
else:
scores.append(split_result['scores']['te'][metric])
if data_type not in score_dfs:
score_dfs[data_type] = {}
score_df = pd.DataFrame({'cancer': cancer, 'target': target,
'type': model_code, 'score': scores})
if metric not in score_dfs[data_type]:
score_dfs[data_type][metric] = score_df
else:
score_dfs[data_type][metric] = pd.concat(
[score_dfs[data_type][metric], score_df], axis=0)
for data_type in data_types:
for metric in metrics:
score_df = score_dfs[data_type][metric].copy()
score_df['model'] = pd.Categorical(score_df['cancer'].str.upper() + ' '
+ score_df['target'], ordered=True)
score_df['type'] = pd.Categorical(
score_df['type'].str.upper(),
categories=[m.upper() for m in model_codes], ordered=True)
score_df['type'] = score_df['type'].cat.remove_unused_categories()
fig, ax = plt.subplots(figsize=(fig_width, fig_height), dpi=fig_dpi)
sns.barplot(x='model', y='score', hue='type', data=score_df, ci='sd',
capsize=0.1, palette=bcolors, errcolor=ecolor,
errwidth=1.75, saturation=1)
# bar_labels = ['***' if p <= 0.0001 else '**' if p <= 0.001 else
# '*' if p <= 0.01 else '$^{ns}$' for p in np.ravel([
# p_adjs[data_type][k] for k in
# score_df['type'].cat.categories.str.lower()])]
# for bar, label in zip(ax.patches, bar_labels):
# ax.annotate(label, (bar.get_x() + bar.get_width() / 2, 1.0),
# ha='center', va='bottom', size=label_fontsize,
# xytext=(0, 1), textcoords='offset points')
for line in ax.get_lines():
x, y = line.get_data()
line.set_data(x, np.clip(y, 0, 1))
sns.stripplot(x='model', y='score', hue='type', data=score_df,
palette=bcolors, edgecolor='black', dodge=True,
jitter=0.15, size=2.5, linewidth=0.8, alpha=0.6,
zorder=2)
# ax.axhline(y=0.6, color='darkgrey', linestyle='--', lw=2, zorder=0)
ax.autoscale(axis='x', enable=None, tight=True)
ax.tick_params(which='major', length=3, width=1.25)
ax.tick_params(axis='x', direction='out', labelsize=x_axis_fontsize,
labelrotation=x_label_rotation, length=3, width=1.25,
pad=0)
ax.set_xlabel(None)
ax.set_ylabel(metric_label[metric], fontsize=y_axis_fontsize,
labelpad=5)
ax.set_yticks(np.arange(0.0, 1.1, 0.2))
ax.get_yaxis().set_major_formatter(ticker.FixedFormatter(
['0', '0.2', '0.4', '0.6', '0.8', '1']))
ax.set_ylim([-0.01, 1.15])
plt.setp(ax.spines.values(), lw=1.25)
ax.spines.right.set_visible(False)
ax.spines.top.set_visible(False)
ax.spines.left.set_bounds(-0.01, 1)
ax.margins(0.01)
ax.grid(True, alpha=0.3, axis='y', which='major')
ax.set_axisbelow(True)
handles, labels = ax.get_legend_handles_labels()
handles, labels = zip(*((h, l) for h, l in zip(handles, labels)
if isinstance(h, BarContainer)))
legend = ax.legend(handles=handles, labels=labels, loc='upper right',
labelspacing=0.25, frameon=False, borderpad=0,
handletextpad=0.25, fontsize=legend_fontsize,
ncol=len(ax.lines), columnspacing=1)
# legend.set_title('Microbiome' if data_type == 'kraken' else
# 'Expression' if data_type == 'htseq' else
# 'Combo', prop={'weight': 'bold',
# 'size': y_axis_fontsize})
legend._legend_box.align = 'right'
fig.tight_layout(pad=0.5, w_pad=0, h_pad=0)
for fmt in args.file_format:
fig.savefig('{}/{}_{}_bar_comp.{}'.format(args.out_dir, data_type,
metric, fmt),
format=fmt, bbox_inches='tight')
score_df.to_csv('{}/{}_{}_bar_comp.tsv'.format(
args.out_dir, data_type, metric), sep='\t', index=False)