-
Notifications
You must be signed in to change notification settings - Fork 0
/
plot-Xu2015.py
304 lines (245 loc) · 12 KB
/
plot-Xu2015.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
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
'''
A Benchmark of Computational CRISPR-Cas9 Guide Design Methods
Jacob Bradford, Dimitri Perrin. 2019.
This script takes the normalised (with scores, generated by normalise-Xu2015.py)
data and plots it to visualise the recall and accuracy of each tool.
Run:
python plotXu2015.py
Input:
- normalised data (with scores) in directory specified in TOOLS_NORMALISED_DATA_DIR
- (process the raw data using normalise-Xu2015.py)
Output:
- PNG file
- see last line of file for details
'''
import matplotlib.pyplot as plt
import argparse, csv, numpy, time, os, re, string
from matplotlib.pyplot import cm, colors
import matplotlib.patches as mpatches
import matplotlib as mpl
import math
''' Config '''
XU_2014_DATA = r"Xu-2015_Is-Efficient.csv"
TOOLS_NORMALISED_DATA_DIR = r"normalised-bauer-lab-updates-exp-with-scores"
PLOT_OUTPUT_DPI = 300
NUMBER_OF_ROWS_IN_PLOT = 2
WIDTH = 20
HEIGHT = 15
PRETTY_NAMES = {
'chopchop' : 'CHOPCHOP',
'gtscan' : 'GT-Scan',
'crispor' : 'CRISPOR',
'crisprera' : 'CRISPR-ERA',
'ctfinder' : 'CT-Finder',
'casdesigner' : 'Cas-Designer',
'mm10crisprdatabase' : 'mm10db',
'sgrnascorer20' : 'sgRNAScorer2',
'guidescan' : 'GuideScan',
'casfinder' : 'CasFinder',
'wucrispr' : 'WU-CRISPR',
'phytocrispex' : 'PhytoCRISP-Ex',
'flashfry' : 'FlashFry',
'sgrnacas9' : 'sgRNAcas9',
'ssc' : 'SSC',
'cctop' : 'CCTop',
'crisprdo' : 'CRISPR-DO',
'tuscan' : 'TUSCAN'
}
PLOT_ORDER = ['chopchop', 'mm10crisprdatabase', 'tuscan', 'sgrnascorer20','crisprdo', 'wucrispr', 'flashfry', 'ssc']
''' Do not modify '''
X_COL_INDEX = 5
Y_COL_INDEX = 6
SEQ_INDEX = 4
IS_EFFICIENT_COL_INDEX = 7
NORMALISED_TOOL_NAME_INDEX = 0
NORMALISED_GUIDE_INDEX = 1
NORMALISED_SCORE_INDEX = 5
GUIDE_START_POS = 10
GUIDE_LEN = 23 #- 3 # ignore the PAM for now
''' Data handlers '''
X_XU_NO_MATCH_DATA = []
Y_XU_NO_MATCH_DATA = []
X_XU_MATCH_DATA = []
Y_XU_MATCH_DATA = []
SEQ_XU_DATA = []
TOOL_DATA = {} # key: tool name, value: array of guides it reported
TOOL_SCORES_DATA = {} # key: tool name, value: array of guides it reported
''' Copied from mm10db target_identification_viaC.py '''
def rc(dna):
complements = string.maketrans('acgtrymkbdhvACGTRYMKBDHV', 'tgcayrkmvhdbTGCAYRKMVHDB')
rcseq = dna.translate(complements)[::-1]
return rcseq
''' Load in the normalised data '''
for root, dirs, files in os.walk(TOOLS_NORMALISED_DATA_DIR):
for file in files:
if '.normalised' not in file:
continue
fileName = os.path.join(TOOLS_NORMALISED_DATA_DIR, file)
print 'Reading: %s' % file
with open(fileName, 'r') as fOpen:
# expecting: crisprera,TACAGCCAGACATAACACCT...,1826,1848,+
# split the file by row, then split by comma
for line in [x.split(',') for x in fOpen.read().split('\n')[:-1]]:
# we want every guide from all tools except chopchop.
# in the case of chopchop, we only want those that are reported
# to be efficient. same for tuscan.
if (line[NORMALISED_TOOL_NAME_INDEX] not in ['chopchop', 'tuscan']) or (line[NORMALISED_TOOL_NAME_INDEX] == 'chopchop' and line[NORMALISED_SCORE_INDEX] == '1.00') or (line[NORMALISED_TOOL_NAME_INDEX] == 'tuscan' and line[NORMALISED_SCORE_INDEX] == '1.0'):
if line[NORMALISED_TOOL_NAME_INDEX] not in TOOL_DATA:
TOOL_DATA[line[NORMALISED_TOOL_NAME_INDEX]] = []
TOOL_SCORES_DATA[line[NORMALISED_TOOL_NAME_INDEX]] = []
TOOL_DATA[line[NORMALISED_TOOL_NAME_INDEX]].append(line[NORMALISED_GUIDE_INDEX])#[:-3])
if line[NORMALISED_SCORE_INDEX] in ['None', 'N/A']:
TOOL_SCORES_DATA[line[NORMALISED_TOOL_NAME_INDEX]].append(None)
else:
TOOL_SCORES_DATA[line[NORMALISED_TOOL_NAME_INDEX]].append(float(line[NORMALISED_SCORE_INDEX]))
''' Repeat for each tool '''
plotCount = 1 # one index
print 'tool,len(X_XU_MATCH_DATA),len(X_XU_NO_MATCH_DATA),MATCH_XU_EFFICIENT_COUNT,MATCH_XU_INEFFICIENT_COUNT'
for tool in PLOT_ORDER:
#effGuideFile = open('%s.eff' % tool, 'w')
#if tool != 'mm10crisprdatabase':
# continue
count = 0
X_XU_NO_MATCH_DATA = []
Y_XU_NO_MATCH_DATA = []
X_XU_MATCH_DATA = []
Y_XU_MATCH_DATA = []
MATCH_XU_EFFICIENT_COUNT = 0
MATCH_XU_INEFFICIENT_COUNT = 0
XU_GUIDE_SEQS_MATCHES_ONLY = []
SCORES = []
XU_IS_EFFICIENT = []
# Load in the Xu 2014 data
with open(XU_2014_DATA, 'r') as fDataRead:
# split the file by row, then split by comma
for line in [x.split(',') for x in fDataRead.read().split('\n')[1:-1]]:
SEQ_XU_DATA.append(line[SEQ_INDEX])
x = float(line[X_COL_INDEX])
y = float(line[Y_COL_INDEX])
indexInNormalisedData = None
XuGuide = line[SEQ_INDEX][GUIDE_START_POS:(GUIDE_START_POS + GUIDE_LEN)][:-3]
forwardMatch = XuGuide in [i[:-3] for i in TOOL_DATA[tool]]
if forwardMatch:
indexInNormalisedData = [i[:-3] for i in TOOL_DATA[tool]].index(XuGuide)
reverseMatch = False
if XuGuide[:2] == 'CC':
reverseMatch = rc(XuGuide) in [i[:-3] for i in TOOL_DATA[tool]]
if indexInNormalisedData is not None:
score = TOOL_SCORES_DATA[tool][indexInNormalisedData]
if score not in [None, 'N/A']:
SCORES.append(score)
if forwardMatch or reverseMatch:
count = count + 1
X_XU_MATCH_DATA.append(x)
Y_XU_MATCH_DATA.append(y)
XU_GUIDE_SEQS_MATCHES_ONLY.append(XuGuide)
if line[IS_EFFICIENT_COL_INDEX] == '1':
XU_IS_EFFICIENT.append(1)
MATCH_XU_EFFICIENT_COUNT += 1
else:
XU_IS_EFFICIENT.append(0)
MATCH_XU_INEFFICIENT_COUNT += 1
else:
X_XU_NO_MATCH_DATA.append(x)
Y_XU_NO_MATCH_DATA.append(y)
if len(SCORES) != len(X_XU_MATCH_DATA):
SCORES=[255.0] * len(X_XU_MATCH_DATA)
ax = plt.subplot(NUMBER_OF_ROWS_IN_PLOT, math.ceil(len(PLOT_ORDER) / NUMBER_OF_ROWS_IN_PLOT), plotCount)
if plotCount in [1,5]:
plt.ylabel('log2 fold-change, KBM-7')
plt.xlabel('log2 fold-change, HL-60')
plt.title(PRETTY_NAMES[tool])
plotCount += 1
if tool == 'sgrnascorer20':
# we want to know how many guides were deemed efficient vs inefficient
# according to Xu in each percentile
ninetyAbove = numpy.percentile(SCORES, 90)
seventyFive = numpy.percentile(SCORES, 75)
fifty = numpy.percentile(SCORES, 50)
# colours
colNinetyAbove = '#d7191c'
colSeventyFive = '#fdae61'
colFifty = '#abd9e9'
colLessFifty = '0.4'
colNinetyAbove = '#00b8ff'
colSeventyFive = '#6aff8d'
colFifty = '#ff7e00'
colLessFifty = '#b20000'
counts = [0, 0, 0, 0] # each index represents a percentile. see below
isEfficientCounts = [[0,0], [0,0], [0,0], [0,0]] # [[inefficient, efficient], ..]
for i in xrange(len(SCORES)):
score = SCORES[i]
# this is so hacky its ridic
if score >= ninetyAbove:
SCORES[i] = colNinetyAbove
counts[0] += 1
# doing this will allow us to know how many guides were deemed by
# Xu as being efficient or inefficient in this particular percentile
isEfficientCounts[0][XU_IS_EFFICIENT[i]] += 1
if score >= seventyFive and score < ninetyAbove:
SCORES[i] = colSeventyFive
counts[1] += 1
isEfficientCounts[1][XU_IS_EFFICIENT[i]] += 1
if score >= fifty and score < seventyFive:
SCORES[i] = colFifty
counts[2] += 1
isEfficientCounts[2][XU_IS_EFFICIENT[i]] += 1
if score < fifty:
SCORES[i] = colLessFifty
counts[3] += 1
isEfficientCounts[3][XU_IS_EFFICIENT[i]] += 1
plt.scatter(X_XU_NO_MATCH_DATA, Y_XU_NO_MATCH_DATA, s=15, marker='.', label='No Match', c='0.85')
plt.scatter(X_XU_MATCH_DATA, Y_XU_MATCH_DATA, s=15, marker='.', label='Match', c=SCORES)
plt.legend(handles=[
mpatches.Patch(color=colNinetyAbove, label='Above 90%'),
mpatches.Patch(color=colSeventyFive, label='75-89%'),
mpatches.Patch(color=colFifty, label='50-74%'),
mpatches.Patch(color=colLessFifty, label='Below 50%'),
mpatches.Patch(color='0.85', label='Not Reported')
], loc='upper left')
print 'sgrnascorer20: %i %i %i %i %i %i' % (counts[0], counts[1], counts[2], counts[3], MATCH_XU_EFFICIENT_COUNT, MATCH_XU_INEFFICIENT_COUNT)
elif tool in ['flashfry', 'casdesigner', 'wucrispr', 'crispor', 'crisprdo', 'ssc']:
minn = min(SCORES) - min(SCORES) * 0.1
maxx = max(SCORES) + max(SCORES) * 0.1
norm = mpl.colors.Normalize(vmin=minn, vmax=maxx)
plt.scatter(X_XU_NO_MATCH_DATA, Y_XU_NO_MATCH_DATA, s=15, marker='.', label='No Match', color='0.85')
plt.scatter(X_XU_MATCH_DATA, Y_XU_MATCH_DATA, s=15, marker='.', label='Match', c=SCORES, cmap=cm.jet_r, norm=norm)
plt.colorbar()
if tool in ['wucrispr', 'crisprdo']:
plt.legend(handles=[
mpatches.Patch(color='0.85', label='Not Reported')
], loc='upper left')
else:
plt.scatter(X_XU_NO_MATCH_DATA, Y_XU_NO_MATCH_DATA, s=15, marker='.', label='No Match', color='0.85')
SCORES = ['#1f78b4' * len(SCORES)]
plt.scatter(X_XU_MATCH_DATA, Y_XU_MATCH_DATA, s=15, marker='.', label='Match', c=['#1f78b4' * len(SCORES)], cmap=cm.copper)
plt.legend(handles=[
mpatches.Patch(color='#1f78b4', label='Accepted'),
mpatches.Patch(color='0.85', label='Rejected')
], loc='upper left')
print '%s\t%i\t%i\t%i\t%i' % (tool, len(X_XU_MATCH_DATA), len(X_XU_NO_MATCH_DATA), MATCH_XU_EFFICIENT_COUNT, MATCH_XU_INEFFICIENT_COUNT)
plt.axhline(0, linestyle='-', color='k', linewidth=0.5)
plt.axvline(0, linestyle='-', color='k', linewidth=0.5)
''' (calculated in excel)
efficient inefficient
min max min max
x log2 fold change, HL60 -5.305412391 -1.001239638 -0.497285084 2.519349582
y log2 fold change, KBM7 -5.498241555 -1.022733387 -0.49447052 2.97226217
'''
# efficient (q3)
ax.hlines(-1.022733387, -5.75, -1.001239638, linestyle='-', color='0.65', linewidth=0.5)
ax.vlines(-1.001239638, -5.75, -1.022733387, linestyle='-', color='0.65', linewidth=0.5)
# inefficient (q1)
ax.hlines(-0.49447052, -0.497285084, 2.75, linestyle='-', color='0.65', linewidth=0.5)
ax.vlines(-0.497285084, -0.49447052, 3, linestyle='-', color='0.65', linewidth=0.5)
plt.text(-5.6, -1.9, 'Efficient', color='0.1', fontsize='small', fontstyle='italic')
plt.text(0.4, 2.4, 'Inefficient', color='0.1', fontsize='small', fontstyle='italic')
effMatchPercent = float(MATCH_XU_EFFICIENT_COUNT) / 731.0 * 100
ineffMatchPercent = float(MATCH_XU_INEFFICIENT_COUNT) / (1169 - 731.0) * 100
plt.text(-5.65, -2.2, '(%3.1f%%)' % (effMatchPercent), color='0.1', fontsize='small', fontstyle='italic')
plt.text(0.5, 2.1, '(%3.1f%%)' % (ineffMatchPercent), color='0.1', fontsize='small', fontstyle='italic')
plt.rcParams.update({'font.size': 13})
fig = plt.gcf()
fig.set_size_inches(WIDTH, HEIGHT)
fig.savefig('plotXu2015.png', dpi=PLOT_OUTPUT_DPI)
#fig.savefig('plotXu2015.eps', format='eps', dpi=PLOT_OUTPUT_DPI)