forked from fschmich/tscan_wrapper
-
Notifications
You must be signed in to change notification settings - Fork 0
/
predicttargets.py
192 lines (167 loc) · 7.3 KB
/
predicttargets.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
#!/usr/bin/env python
# Run TargetScan's context score script (targetscan_60_context_scores.pl) on
# the output of TargetScan's base script (targetscan_60.pl). The script takes
# as input the id and sequence of an siRNA, as well as the directory of seeds
# scanned with TargetScan's base script and a translation file from transcripts
# to genes, and outputs TargetScan context score predictions to an output
# folder. Remember, we cannot use aligned UTRs, since siRNAs are not endogenous
# entities.
#
# Authors: Fabian Schmich ([email protected]) and Mason Victors
#
# Libraries
import sys
import os.path
import optparse
import tempfile
import subprocess
import pandas as pd
from numpy import mean
# Paths
PROC = subprocess.Popen(['bash',
'-c',
'which targetscan_60_context_scores.pl'],
stdout=subprocess.PIPE, stderr=subprocess.PIPE,
stdin=subprocess.PIPE)
STDOUT, STDERR = PROC.communicate()
CS_SCRIPT = STDOUT.strip()
CS_SCRIPT_DIR = os.path.dirname(CS_SCRIPT)
class Target:
"Class to represent a siRNA target gene"
def __init__(self, id, name=None, synonyms=[],
scores=[], transcripts=[]):
self.id = id
self.name = name
self.synonyms = synonyms
self.scores = scores
self.transcripts = transcripts
def __hash__(self):
return hash(self.id)
def __eq__(self, other):
return self.id == other.id
def __repr__(self):
if self.name is not None:
return "\t".join([str(self.id),
self.name,
';'.join(self.synonyms)])
else:
return str(self.id)
def strextd(self):
return "\t".join([str(self.id),
self.name,
";".join(self.synonyms),
";".join(self.transcripts),
";".join([str(e) for e in self.scores]),
str(mean(self.scores))])
# Takes an siRNAs and produces the input files required by context_score
# script, "miRNA file" and "PredictedTargets" file by matching the sRNA
# seed to pre-scanned seeds.
def prepare(seq, targetscan_60_outdir, species="9606"):
# Mature sequence file
mature_sequence_file = tempfile.NamedTemporaryFile("w",
prefix="seq_",
delete=False)
mature_sequence_file.write("\t".join(["miRNA_family_ID",
"Species_ID",
"MiRBase_ID",
"Mature_sequence"]) + "\n")
mature_sequence_file.write("\t".join([seq,
str(species),
seq,
seq]) + "\n")
mature_sequence_file.close()
# Seed predictions file
ts_predictions_file = tempfile.NamedTemporaryFile("w",
prefix="targets_",
delete=False)
seed_filename = os.path.join(targetscan_60_outdir,
'tscan.%s.tsv' % seq)
with open(seed_filename, "r") as seedsf:
ts_predictions_file.write(seedsf.readline()) # header
for line in seedsf:
lsp = line.split("\t")
lsp[1] = seq
ts_predictions_file.write("\t".join(lsp))
ts_predictions_file.close()
return(mature_sequence_file.name, ts_predictions_file.name)
def predict(mat_seq_file_name, ts_pred_file_name, utr_file,
ta_sps_file_name):
contextplus_score_file = tempfile.NamedTemporaryFile("w",
prefix="cps_",
delete=True)
cps_fname = contextplus_score_file.name
contextplus_score_file.close()
olddir = os.getcwd()
#os.chdir(CS_SCRIPT_DIR)
#ta_sps_fname = os.path.join(os.getcwd(), 'TA_SPS_by_seed_region.txt')
from subprocess import call
call([CS_SCRIPT, mat_seq_file_name, utr_file,
ts_pred_file_name, cps_fname, ta_sps_file_name])
os.chdir(olddir)
return cps_fname
def get_translation_dict(ref_seq_file):
with open(ref_seq_file, "r") as translf:
tr = dict()
for line in translf:
lsp = line.split("\t")
lsp = [el.strip() for el in lsp]
if lsp[0] not in tr:
tr[lsp[0]] = [[lsp[1], lsp[2], lsp[3]]]
else:
tr[lsp[0]].append([lsp[1], lsp[2], lsp[3]])
return tr
def get_transcript_dict(csfile):
df = pd.DataFrame.from_csv(csfile, sep='\t', index_col=False)
if df.dtypes['context+ score'] == 'O':
df = df[df['context+ score'] != 'too_close']
transcripts = {k: map(float, v['context+ score'].values)
for k, v in df.groupby('Gene ID')}
return transcripts
def get_targets(transcript_dict, translation_dict):
# Translate
targets = []
for trkey in transcript_dict.keys():
trname = trkey
if "." in trname:
trname = trname[:trname.index(".")] # Strip everything after "."
if trname in translation_dict:
for [gid, gn, syn] in translation_dict[trname]:
curgene = Target(id=gid, name=gn, synonyms=syn.split('|'))
if curgene in targets:
g_ind = targets.index(curgene)
targets[g_ind].scores.append(sum(transcript_dict[trkey]))
targets[g_ind].transcripts.append(trname)
else:
curgene.scores = [sum(transcript_dict[trkey])]
curgene.transcripts = [trname]
targets.append(curgene)
else:
pass
#print("WARN: cannot translate transcript %s" % trname)
target_frame = pd.DataFrame([target.strextd().split('\t')
for target in targets],
columns=['GeneID', 'GeneName',
'GeneSynonyms',
'Transcripts', 'CPS',
'CPSmean'])
return target_frame
def write_target_frame(seq, tscan_outdir, utr_file, ta_sps_file_name,
outdir, ref_seq_file=None,
translation_dict=None):
print("Processing Seed Sequence: %s" % seq)
out_fname = os.path.join(outdir, '%s.tsv' % seq)
if os.path.isfile(out_fname):
return None
mat_seq_file, ts_pred_file = prepare(seq, tscan_outdir)
# Predict context plus scores
contextplus_score_file = predict(mat_seq_file, ts_pred_file, utr_file,
ta_sps_file_name)
transcript_dict = get_transcript_dict(contextplus_score_file)
if translation_dict is None:
translation_dict = get_translation_dict(ref_seq_file)
target_frame = get_targets(transcript_dict, translation_dict)
target_frame.to_csv(out_fname, sep='\t', index=False)
for f in [mat_seq_file, ts_pred_file, contextplus_score_file]:
os.remove(f)
return target_frame[['GeneID', 'GeneName', 'GeneSynonyms']].drop_duplicates()