-
Notifications
You must be signed in to change notification settings - Fork 3
/
crisot_modules.py
275 lines (246 loc) · 13.4 KB
/
crisot_modules.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
import pandas as pd
import numpy as np
import os
import pickle
import time
pwd = os.path.dirname(os.path.realpath(__file__))
def load_pkl(pkl):
with open(pkl, 'rb') as f:
data = pickle.load(f)
return data
param_df, a_b, bins, weights = load_pkl(os.path.join(pwd, 'models/crisot_score_param.pkl'))
class CRISOT:
def __init__(self, param, ref_genome, a_b=a_b, cutoff=None, opti_th=None, prob_weight=weights, bins=bins):
self.feat_dict = {}
for key in param.index:
for i in range(20):
self.feat_dict['Pos' + str(i + 1) + '_' + key] = param.loc[key, :].values[i]
self.ref_genome = ref_genome
if a_b == None:
a_b = [1,0]
self.a_b = a_b
if cutoff == None:
cutoff = 0.
self.cutoff = cutoff
if opti_th == None:
opti_th = 0.8
self.opti_th = opti_th
self.prob_weight = prob_weight
self.bins = bins
self.proj_t = time.time()
def single_score_(self, on_seq, off_seq):
y_pred = np.array([self.feat_dict['Pos' + str(j + 1) + '_' + on_seq[j] + off_seq[j]] for j in range(20)]).sum()
y_pred = y_pred * self.a_b[0] + self.a_b[1]
if y_pred > 1:
y_pred = 1.
elif y_pred < 0:
y_pred = 0.
return y_pred
def score(self, data_path=None, data_df=None, On='On', Off='Off', Active=None):
if data_df is not None:
data_set = data_df
else:
data_set = pd.read_csv(data_path, sep=",", header=0, index_col=None)
ont = data_set.loc[:, On].values
offt = data_set.loc[:, Off].values
y_pred = np.array([self.single_score_(ont[i], offt[i]) for i in range(ont.shape[0])])
if Active == None:
return y_pred
else:
y_ori = data_set.loc[:, Active].values
return y_ori, y_pred
def score_bin_(self, y_pred):
y_pred = np.array(y_pred)
if y_pred.shape[0] != 0:
y_df = pd.DataFrame(y_pred.reshape(-1,1), columns=['CRISOT-Score'])
y_count = y_df['CRISOT-Score'].value_counts(bins=self.bins, sort=False).values
else:
y_count = np.array([0] * (len(self.bins)-1))
return y_count
def single_aggre_(self, y_pred, out_cnt=True):
cnt = self.score_bin_(y_pred)
aggre = (cnt * self.prob_weight).sum()
if out_cnt:
return np.append(cnt[-4:], aggre)
else:
return aggre
def single_spec_(self, y_pred):
aggre = self.single_aggre_(y_pred, out_cnt=False)
spec = 10 / (10 + aggre)
return spec
def spec(self, data_path=None, data_df=None, On='On', Off='Off', target=None, out_df=False):
if data_df is not None:
data_set = data_df
else:
data_set = pd.read_csv(data_path, sep=",", header=0, index_col=None)
offt = data_set.loc[:, Off].values
if target is not None:
assert len(target) == 23, 'target sequence must have 23 nt'
assert data_set.loc[offt == target, :].shape[0] > 0, 'No sequence match the target sequence'
data_set = pd.concat([data_set.loc[offt == target, :], data_set.loc[offt != target, :]])
y_pred = self.score(data_df=data_set, On=On, Off=Off, Active=None)
spec = self.single_spec_(y_pred[1:])
if out_df:
data_set['CRISOT-Score'] = y_pred
return spec, data_set
else:
return spec
def CasoffinderSpec_(self, sgrna, target, out_df=False, offtar_search=None, mm=6, dev='C'):
if offtar_search is not None:
offtar_search = offtar_search
else:
offtar_search = os.path.join(pwd, 'script/casoffinder_genome.sh')
if os.path.exists(f'.temp_{self.proj_t}_casoffinder.out'):
os.system(f'rm .temp_{self.proj_t}_casoffinder.out')
if os.path.exists(f'.temp_{self.proj_t}_casoffinder.in'):
os.system(f'rm .temp_{self.proj_t}_casoffinder.in')
os.system("sh {} {} {} {} {} {}".format(offtar_search, sgrna[:20], self.ref_genome, mm, dev, self.proj_t))
data_set = pd.read_csv(f'.temp_{self.proj_t}_casoffinder.out', sep="\t", header=None, index_col=None)
offt = data_set.loc[:, 3].values
offt = np.array([str.upper(t) for t in offt])
data_set.loc[:, 3] = offt
data_set = data_set[-data_set[3].str.contains('N|R|W|M|V|Y|K|D|S|J')]
data_set.drop_duplicates([1,2,3], inplace=True)
if data_set[data_set[3].str.contains(target[:20])].shape[0] == 0:
data_set = data_set.append(pd.Series([data_set.iloc[0, 0], np.nan, np.nan, target, np.nan, np.nan]), ignore_index=True)
data_set = pd.concat(
[data_set[data_set[3].str.contains(target[:20])], data_set[-data_set[3].str.contains(target[:20])]])
if out_df:
spec, out_dset = self.spec(data_df=data_set, On=0, Off=3, out_df=out_df)
else:
spec = self.spec(data_df=data_set, On=0, Off=3)
os.system(f"rm .temp_{self.proj_t}_casoffinder.out")
if out_df:
return spec, out_dset
else:
return spec
def opti(self, target, opti_type=None, ref=None, offtar_search=None, mm=6, dev='G0'):
# opti_type: how to mutate the sgRNA? default is 3 types of mutations for each position
if opti_type == None:
opti_pos = []
opti_seq = []
for p in range(1, 21):
opti_pos = np.append(opti_pos, np.array([p, p, p, p], dtype='int'))
opti_seq = np.append(opti_seq, ['A', 'C', 'G', 'T'])
opti_nt = pd.DataFrame()
opti_nt['Pos'] = opti_pos
opti_nt['nt'] = opti_seq
elif type(opti_type) == tuple:
opti_pos = []
opti_seq = []
for p in range(opti_type[0], opti_type[1]):
opti_pos = np.append(opti_pos, np.array([p, p, p, p], dtype='int'))
opti_seq = np.append(opti_seq, ['A', 'C', 'G', 'T'])
opti_nt = pd.DataFrame()
opti_nt['Pos'] = opti_pos
opti_nt['nt'] = opti_seq
elif (type(opti_type) == np.ndarray) or (type(opti_type) == list) or (type(opti_type) == range):
opti_pos = []
opti_seq = []
for p in opti_type:
opti_pos = np.append(opti_pos, np.array([p, p, p, p], dtype='int'))
opti_seq = np.append(opti_seq, ['A', 'C', 'G', 'T'])
opti_nt = pd.DataFrame()
opti_nt['Pos'] = opti_pos
opti_nt['nt'] = opti_seq
elif type(opti_type) == pd.core.frame.DataFrame:
opti_nt = opti_type
else:
opti_nt = pd.read_csv(opti_type, header=0, index_col=0)
results = []
# calculate WT spec score
spec, out_df = self.CasoffinderSpec_(target, target, out_df=True, offtar_search=offtar_search, mm=mm, dev=dev)
cnt_results = self.score_bin_(out_df.loc[:, 'CRISOT-Score'].values[1:])[-5:]
cnt_results = np.append(cnt_results[:3], cnt_results[3:].sum())
if ref is not None:
ref_merge = ref.loc[:, ['Offtarget_Sequence', 'GUIDE-Seq Reads']]
ref_merge.columns = [3, 'Reads']
out_merge = pd.merge(out_df.iloc[1:,:], ref_merge, how='left', on=[3], sort=False)
out_merge.fillna(0, inplace=True)
out_merge = out_merge.loc[out_merge.loc[:,'Reads'] > 0, :]
if out_merge.shape[0] != 0:
out_merge.to_csv('{}_ref_match.tsv'.format(target), sep='\t')
spec_ref = self.single_spec_(out_merge.loc[:, 'CRISOT-Score'].values[1:], out_cnt=False)
else:
spec_ref = 1.0
results.append([target, target, 'WT', out_df.loc[:, 'CRISOT-Score'].values[0]] + list(cnt_results) + [spec, spec_ref])
else:
results.append([target, target, 'WT', out_df.loc[:, 'CRISOT-Score'].values[0]] + list(cnt_results) + [spec])
# calculate spec scores of modified sgRNAs
for i in range(opti_nt.shape[0]):
pos = int(opti_nt.loc[i, 'Pos'])
nt = opti_nt.loc[i, 'nt']
if target[pos - 1] != nt:
new_sgrna = target[:pos - 1] + nt + target[pos:]
if self.single_score_(new_sgrna, target) > self.opti_th:
spec, out_df = self.CasoffinderSpec_(new_sgrna, target, out_df=True, offtar_search=offtar_search, mm=mm, dev=dev)
cnt_results = self.score_bin_(out_df.loc[:, 'CRISOT-Score'].values[1:])[-5:]
cnt_results = np.append(cnt_results[:3], cnt_results[3:].sum())
if ref is not None:
out_merge = pd.merge(out_df.iloc[1:,:], ref_merge, how='left', on=[3], sort=False)
out_merge.fillna(0, inplace=True)
out_merge = out_merge.loc[out_merge.loc[:,'Reads'] > 0, :]
if out_merge.shape[0] != 0:
spec_ref = self.single_spec_(out_merge.loc[:, 'CRISOT-Score'].values[1:], out_cnt=False)
else:
spec_ref = 1.0
if ref is not None:
results.append([new_sgrna, target, '{}{}>{}'.format(target[pos - 1], pos, nt), out_df.loc[:, 'CRISOT-Score'].values[0]] + list(cnt_results) + [spec, spec_ref])
else:
results.append([new_sgrna, target, '{}{}>{}'.format(target[pos - 1], pos, nt), out_df.loc[:, 'CRISOT-Score'].values[0]] + list(cnt_results) + [spec])
results = np.array(results)
if results.shape[0] == 1:
results = np.vstack([results, np.array([[np.nan for i in range(results.shape[1])]], dtype=object)])
results[1, 1:3] = np.array([target, 'Optimization unavailable'])
if ref is not None:
results_df = pd.DataFrame(results, columns=['sgRNA', 'Target', 'Mutation', 'CRISOT-Score', '(0.65, 0.7]', '(0.7, 0.75]',
'(0.75, 0.8]', '(0.8, 1.0]', 'CRISOT-Spec', 'ref_Spec'])
else:
results_df = pd.DataFrame(results, columns=['sgRNA', 'Target', 'Mutation', 'CRISOT-Score', '(0.65, 0.7]', '(0.7, 0.75]',
'(0.75, 0.8]', '(0.8, 1.0]', 'CRISOT-Spec'])
results_df.iloc[:, 3:] = np.array(results_df.iloc[:, 3:].values, dtype='float')
try:
results_df.iloc[:, 4:8] = np.array(results_df.iloc[:, 4:8].values, dtype='int')
except:
results_df.iloc[0, 4:8] = np.array(results_df.iloc[0, 4:8].values, dtype='int')
results_df['delta_Spec'] = results_df.loc[:, 'CRISOT-Spec'].values - results_df.loc[0, 'CRISOT-Spec']
results_df['rk'] = np.array([0] + [1] * (results_df.shape[0]-1))
results_out = results_df.sort_values(by=['rk', 'delta_Spec'], ascending=[True, False])
results_out.index = np.arange(results_out.shape[0])
return results_out.iloc[:, :-1]
def rescore(self, targets, ref=None, offtar_search=None, mm=6, dev='G0'):
dataread = pd.DataFrame(np.array([np.arange(len(targets)) + 1, targets]).T, columns=['Ori_Rank', 'Target sequence'])
results = []
# calculate WT spec score
for i in range(len(targets)):
target = targets[i]
spec, out_df = self.CasoffinderSpec_(target, target, out_df=True, offtar_search=offtar_search, mm=mm, dev=dev)
cnt_results = self.score_bin_(out_df.loc[:, 'CRISOT-Score'].values[1:])[-5:]
cnt_results = np.append(cnt_results[:3], cnt_results[3:].sum())
if ref is not None:
ref_merge = ref.loc[:, ['Offtarget_Sequence', 'GUIDE-Seq Reads']]
ref_merge.columns = [3, 'Reads']
out_merge = pd.merge(out_df.iloc[1:,:], ref_merge, how='left', on=[3], sort=False)
out_merge.fillna(0, inplace=True)
out_merge = out_merge.loc[out_merge.loc[:,'Reads'] > 0, :]
if out_merge.shape[0] != 0:
# out_merge.to_csv('{}_ref_match.tsv'.format(target), sep='\t')
spec_ref = self.single_score_(out_merge.loc[:, 'CRISOT-Score'].values[1:], out_cnt=False)
else:
spec_ref = 1.0
if ref is not None:
results.append(np.append(np.append(out_df.loc[:, 'CRISOT-Score'].values[0], cnt_results), [spec, spec_ref]))
results_df = pd.DataFrame(results, columns=['CRISOT-Score', '(0.65, 0.7]', '(0.7, 0.75]',
'(0.75, 0.8]', '(0.8, 1.0]', 'CRISOT-Spec', 'ref_Spec'])
else:
results.append(np.append(np.append(out_df.loc[:, 'CRISOT-Score'].values[0], cnt_results), [spec]))
results_df = pd.DataFrame(results, columns=['CRISOT-Score', '(0.65, 0.7]', '(0.7, 0.75]',
'(0.75, 0.8]', '(0.8, 1.0]', 'CRISOT-Spec'])
results_df.iloc[:, :] = np.array(results_df.iloc[:, :].values, dtype='float')
results_df = pd.concat([dataread, results_df], axis=1)
rk = np.array(results_df.loc[:, 'CRISOT-Score'].values > self.opti_th, dtype=int)
results_df['CRISOT_rank'] = rk
results_out = results_df.sort_values(by=['CRISOT_rank', 'CRISOT-Spec'], ascending=[False, False])
results_out.index = np.arange(results_out.shape[0])
results_out.loc[:, 'CRISOT_rank'] = np.arange(1, results_out.shape[0]+1)
return results_out