vardetect-vc 0.0.4 → 0.0.5
Sign up to get free protection for your applications and to get access to all the features.
- data/bin/vardetect_vc +46 -0
- data/bin/vardetect_vc_batch +9 -0
- data/lib/vardetect-vc/call_combine.rb +367 -0
- data/lib/vardetect-vc/call_prepare.rb +122 -0
- data/lib/vardetect-vc/call_snv.rb +493 -0
- data/lib/vardetect-vc/lib.rb +28 -0
- data/lib/vardetect-vc/version.rb +1 -1
- metadata +10 -2
data/bin/vardetect_vc
ADDED
@@ -0,0 +1,46 @@
|
|
1
|
+
#!/usr/bin/env ruby
|
2
|
+
require 'rubygems'
|
3
|
+
require 'vardetect-vc'
|
4
|
+
require "vardetect-vc/lib"
|
5
|
+
|
6
|
+
|
7
|
+
sample_id = 'G3418'
|
8
|
+
working = 'vcqa'
|
9
|
+
|
10
|
+
path = '/Users/soup/Desktop/chula/data'
|
11
|
+
ref_file = File.join(path,'ref','hg19.fa')
|
12
|
+
|
13
|
+
|
14
|
+
snp_array_file = File.join(path,working,"#{sample_id}_gt_report.txt")
|
15
|
+
sam_tool_file = File.join(path,working,"#{sample_id}_SNP_Indel_ANNO.csv")
|
16
|
+
sam_file = File.join(path,working,'bam',"#{sample_id}.remdup.uniqMap.TS.bam")
|
17
|
+
vc_raw_file = File.join(path,working,'snv-vc',"#{sample_id}_vc_raw_snv.tsv")
|
18
|
+
|
19
|
+
|
20
|
+
st_file = File.join(path,working,sample_id,'st_snv.tsv')
|
21
|
+
vc_file = File.join(path,working,sample_id,'vc_snv.csv')
|
22
|
+
snp_file = File.join(path,working,sample_id,'snpar_snv.csv')
|
23
|
+
|
24
|
+
st_ft_file = File.join(path,working,sample_id,'st_filtered_snv.csv')
|
25
|
+
vc_ft_file = File.join(path,working,sample_id,'vc_filtered_snv.csv')
|
26
|
+
snp_ft_file = File.join(path,working,sample_id,'snpar_filtered_snv.csv')
|
27
|
+
|
28
|
+
|
29
|
+
com_vc_file = File.join(path,working,sample_id,'vc_compared.csv')
|
30
|
+
com_st_file = File.join(path,working,sample_id,'st_compared.csv')
|
31
|
+
com_all_file = File.join(path,working,sample_id,'all_compared.csv')
|
32
|
+
|
33
|
+
combine_all_file = File.join(path,working,sample_id,'combine_all.csv')
|
34
|
+
analysed_file = File.join(path,working,sample_id,'summary.csv')
|
35
|
+
|
36
|
+
# 1. call vc
|
37
|
+
params = "snv -ref #{ref_file} -sam #{sam_file} -vc_out #{vc_raw_file} -core 8"
|
38
|
+
# params = "snv-range -ref #{ref_file} -sam #{sam_file} -vc_out #{vc_raw_file} -core 8 -chr 1 -start 168665645 -stop 170706648 -debug true"
|
39
|
+
params = params.split
|
40
|
+
# 156553003
|
41
|
+
#
|
42
|
+
#
|
43
|
+
# 170706648
|
44
|
+
|
45
|
+
Vardetect::Vc.hi
|
46
|
+
Vardetect::Vc.exec params
|
@@ -0,0 +1,367 @@
|
|
1
|
+
module Vardetect
|
2
|
+
module Vc
|
3
|
+
|
4
|
+
|
5
|
+
def self.call_combine_all params
|
6
|
+
# ====================================================================================================
|
7
|
+
# 4. All comparison
|
8
|
+
# ====================================================================================================
|
9
|
+
|
10
|
+
map_vc = []
|
11
|
+
map_st = {}
|
12
|
+
map_snp = {}
|
13
|
+
|
14
|
+
|
15
|
+
vc_file =params['vc_out']
|
16
|
+
st_file =params['st_out']
|
17
|
+
snp_file =params['snp_out']
|
18
|
+
|
19
|
+
com_out =params['com_out']
|
20
|
+
|
21
|
+
f_st = File.open(st_file,'r')
|
22
|
+
f_vc = File.open(vc_file,'r')
|
23
|
+
f_snp = File.open(snp_file,'r')
|
24
|
+
|
25
|
+
com_output = File.open(com_out,'w')
|
26
|
+
|
27
|
+
puts "Indexing SNP Array SNV"
|
28
|
+
while str = f_snp.gets
|
29
|
+
v = str.strip.split(',')
|
30
|
+
map_snp[v[0]] = v
|
31
|
+
end
|
32
|
+
|
33
|
+
puts map_snp.keys[0..10]
|
34
|
+
|
35
|
+
puts "Indexing SamTools SNV"
|
36
|
+
while str = f_st.gets
|
37
|
+
v = str.strip.split(',')
|
38
|
+
map_st[v[0]] = v
|
39
|
+
end
|
40
|
+
|
41
|
+
puts map_st.keys[0..10]
|
42
|
+
|
43
|
+
puts "Indexing VC SNV"
|
44
|
+
while str = f_vc.gets
|
45
|
+
v = str.split(',')
|
46
|
+
|
47
|
+
|
48
|
+
|
49
|
+
# if (v[2]==v[3] and alt.to_f/dep > 0.1) or v[2]!=v[3]
|
50
|
+
# map_vc<<v
|
51
|
+
# end
|
52
|
+
#
|
53
|
+
# if (v[2]==v[3] and alt.to_f/dep > 0.1) or v[2]!=v[3]
|
54
|
+
map_vc<<v
|
55
|
+
# end
|
56
|
+
|
57
|
+
|
58
|
+
end
|
59
|
+
|
60
|
+
puts map_vc.size
|
61
|
+
|
62
|
+
# puts map_vc[0..10]
|
63
|
+
|
64
|
+
|
65
|
+
|
66
|
+
|
67
|
+
|
68
|
+
# map_vc.sort!{|a,b| a[1].to_i<=>b[1].to_i }
|
69
|
+
|
70
|
+
all=0
|
71
|
+
vc_count=0
|
72
|
+
st_count=0
|
73
|
+
vc_het_rate=0
|
74
|
+
st_het_rate=0
|
75
|
+
final = []
|
76
|
+
|
77
|
+
|
78
|
+
for i in map_vc
|
79
|
+
pos = i[0]
|
80
|
+
# puts pos
|
81
|
+
if snp = map_snp[pos]
|
82
|
+
|
83
|
+
|
84
|
+
vc_alt = i[5].to_i
|
85
|
+
vc_dep = i[4].to_i+vc_alt
|
86
|
+
|
87
|
+
vc_alt_a = '-'
|
88
|
+
vc_het = 0
|
89
|
+
|
90
|
+
|
91
|
+
snp_rs = snp[1]
|
92
|
+
snp_a1 = snp[2]
|
93
|
+
snp_a2 = snp[3]
|
94
|
+
|
95
|
+
snp_a = "#{snp_a1}/#{snp_a2}"
|
96
|
+
snp_het = 0
|
97
|
+
snp_het = 1 if snp_a1!=snp_a2
|
98
|
+
snp_gc = snp[4]
|
99
|
+
snp_clust = snp[5]
|
100
|
+
snp_cmp = 1
|
101
|
+
|
102
|
+
snp_check = false
|
103
|
+
if snp_a.strip !='-/-'
|
104
|
+
all+=1
|
105
|
+
snp_check = true
|
106
|
+
end
|
107
|
+
st_check = false
|
108
|
+
|
109
|
+
vc_het_t = 'hom'
|
110
|
+
snp_het_t = 'hom'
|
111
|
+
snp_het_t = 'het' if snp_a1!=snp_a2
|
112
|
+
|
113
|
+
vc_het_ratio = vc_alt.to_f/vc_dep
|
114
|
+
if vc_het_ratio >= 0.15
|
115
|
+
vc_alt_a = i[3]
|
116
|
+
vc_het = 1
|
117
|
+
vc_het_t = 'het'
|
118
|
+
end
|
119
|
+
|
120
|
+
str = "#{i[0]},#{i[1]},#{i[2]},#{i[2]}/#{vc_alt_a},#{vc_het_t},#{vc_dep},#{vc_alt},#{vc_het_ratio},"
|
121
|
+
str += "#{snp_rs},#{snp_a},#{snp_het_t},#{snp_gc},#{snp_clust},"
|
122
|
+
|
123
|
+
|
124
|
+
|
125
|
+
|
126
|
+
if st = map_st[pos]
|
127
|
+
st_het = 0
|
128
|
+
st_het = 1 if st[5] =='het'
|
129
|
+
str += st[4..-1].join(",")
|
130
|
+
if st_het==snp_het and snp_check
|
131
|
+
st_het_rate+=1
|
132
|
+
st_check =true
|
133
|
+
end
|
134
|
+
st_count+=1
|
135
|
+
end
|
136
|
+
vc_check = false
|
137
|
+
vc_check = true if vc_het==snp_het and snp_check
|
138
|
+
|
139
|
+
if snp_check and vc_check
|
140
|
+
vc_het_rate+=1
|
141
|
+
end
|
142
|
+
vc_count+=1
|
143
|
+
|
144
|
+
cmp = [snp_check,vc_check,st_check,str]
|
145
|
+
|
146
|
+
# final << cmp
|
147
|
+
com_output.puts cmp.join(',')
|
148
|
+
|
149
|
+
|
150
|
+
end
|
151
|
+
end
|
152
|
+
|
153
|
+
|
154
|
+
puts "All #{all} is not -/-"
|
155
|
+
|
156
|
+
puts "VC Hetero Rate : #{vc_het_rate}"
|
157
|
+
|
158
|
+
puts "ST Hetero Rate : #{st_het_rate}"
|
159
|
+
|
160
|
+
f_st.close
|
161
|
+
f_vc.close
|
162
|
+
f_snp.close
|
163
|
+
com_output.close
|
164
|
+
|
165
|
+
|
166
|
+
end
|
167
|
+
|
168
|
+
|
169
|
+
def self.analyse_combine params
|
170
|
+
|
171
|
+
|
172
|
+
# index columns
|
173
|
+
idx = {}
|
174
|
+
%w{snp vc st pos ref vc_a1 vc_a vc_het vc_dep vc_alt vc_r snp_ann snp_a snp_het snp_gc snp_clus st_ref st_a2 st_qua st_dep st_alt st_pos}.each_with_index do |i,index|
|
175
|
+
idx[i.to_sym] = index
|
176
|
+
end
|
177
|
+
|
178
|
+
# constant
|
179
|
+
snp_filter = 0.75
|
180
|
+
alt_ratio = 0.1
|
181
|
+
|
182
|
+
|
183
|
+
gcid = idx[:snp_gc]
|
184
|
+
hid = idx[:snp_het]
|
185
|
+
depid= idx[:vc_dep]
|
186
|
+
|
187
|
+
|
188
|
+
combine_file =params['com_out']
|
189
|
+
analyse_file =params['out']
|
190
|
+
f_list = File.open(combine_file,'r')
|
191
|
+
out = File.open(analyse_file,'w')
|
192
|
+
|
193
|
+
all_list = []
|
194
|
+
list = []
|
195
|
+
|
196
|
+
het_list = []
|
197
|
+
hom_list = []
|
198
|
+
|
199
|
+
|
200
|
+
# filter all by gc score snp array
|
201
|
+
while str = f_list.gets
|
202
|
+
# puts str
|
203
|
+
v = str.strip.split(',')
|
204
|
+
all_list<<v
|
205
|
+
if v[gcid].to_f>snp_filter
|
206
|
+
list<<v
|
207
|
+
end
|
208
|
+
end
|
209
|
+
|
210
|
+
# filter set of snp array 's het hom
|
211
|
+
for i in list
|
212
|
+
if i[hid]=='het'
|
213
|
+
het_list<<i
|
214
|
+
else
|
215
|
+
hom_list<<i
|
216
|
+
end
|
217
|
+
end
|
218
|
+
|
219
|
+
|
220
|
+
|
221
|
+
out.puts "SNP Array overlap data "
|
222
|
+
out.puts "Filter GC score at,#{snp_filter}"
|
223
|
+
out.puts "Total snv,#{all_list.size}"
|
224
|
+
out.puts "Total snv not pass filter,#{all_list.size-list.size}"
|
225
|
+
out.puts "Total hetero snv,#{het_list.size}"
|
226
|
+
out.puts "Total homo snv,#{hom_list.size}"
|
227
|
+
|
228
|
+
|
229
|
+
|
230
|
+
vc_het_het = []
|
231
|
+
vc_het_hom = []
|
232
|
+
vc_hom_hom = []
|
233
|
+
vc_hom_het = []
|
234
|
+
|
235
|
+
vchid = getidx :vc_het
|
236
|
+
vcrid = getidx :vc_r
|
237
|
+
[vchid]=='het' or
|
238
|
+
for i in list
|
239
|
+
|
240
|
+
if i[hid]=='het'
|
241
|
+
if i[vcrid].to_f >= alt_ratio
|
242
|
+
vc_het_het<<i
|
243
|
+
else
|
244
|
+
vc_het_hom<<i
|
245
|
+
end
|
246
|
+
else
|
247
|
+
if i[vcrid].to_f >= alt_ratio
|
248
|
+
vc_hom_het<<i
|
249
|
+
else
|
250
|
+
vc_hom_hom<<i
|
251
|
+
end
|
252
|
+
end
|
253
|
+
end
|
254
|
+
|
255
|
+
|
256
|
+
# matrix=[[vc_het_het.size,vc_het_hom.size],[vc_hom_hom.size,vc_hom_het.size]]
|
257
|
+
|
258
|
+
|
259
|
+
|
260
|
+
require 'erb'
|
261
|
+
outtmp = File.open(analyse_file+".html",'w')
|
262
|
+
|
263
|
+
template = ERB.new(File.open('index.html.erb','r').read)
|
264
|
+
|
265
|
+
|
266
|
+
outtmp.puts template.result(binding)
|
267
|
+
|
268
|
+
outtmp.close
|
269
|
+
|
270
|
+
`open #{analyse_file+".html"}`
|
271
|
+
|
272
|
+
|
273
|
+
|
274
|
+
out.close
|
275
|
+
f_list.close
|
276
|
+
end
|
277
|
+
|
278
|
+
|
279
|
+
def self.getidx id
|
280
|
+
|
281
|
+
idx = {}
|
282
|
+
%w{snp vc st pos ref vc_a1 vc_a vc_het vc_dep vc_alt vc_r snp_ann snp_a snp_het snp_gc snp_clus st_ref st_a2 st_qua st_dep st_alt st_pos}.each_with_index do |i,index|
|
283
|
+
idx[i.to_sym] = index
|
284
|
+
end
|
285
|
+
return idx[id]
|
286
|
+
|
287
|
+
end
|
288
|
+
|
289
|
+
|
290
|
+
def self.sampling list
|
291
|
+
|
292
|
+
gcid = getidx :snp_gc
|
293
|
+
depid = getidx :vc_dep
|
294
|
+
vcrid = getidx :vc_r
|
295
|
+
|
296
|
+
|
297
|
+
# =====================================================================
|
298
|
+
# sampling gc score value from all data set
|
299
|
+
gc_bin = []
|
300
|
+
gc_range = 100
|
301
|
+
gc_max = 1
|
302
|
+
for i in list
|
303
|
+
j = i[gcid].to_f
|
304
|
+
k = (j * gc_range / gc_max).to_i
|
305
|
+
gc_bin[k]=0 unless gc_bin[k]
|
306
|
+
gc_bin[k]+=1
|
307
|
+
end
|
308
|
+
|
309
|
+
# =====================================================================
|
310
|
+
# sampling dep score value from all data set
|
311
|
+
d_bin = []
|
312
|
+
d_range = 100
|
313
|
+
d_max = 500
|
314
|
+
|
315
|
+
for i in list
|
316
|
+
j = i[depid].to_i
|
317
|
+
k = (j * d_range / d_max).to_i
|
318
|
+
d_bin[k]=0 unless d_bin[k]
|
319
|
+
d_bin[k]+=1
|
320
|
+
end
|
321
|
+
|
322
|
+
# =====================================================================
|
323
|
+
# sampling gc score value from all data set
|
324
|
+
|
325
|
+
alt_bin = []
|
326
|
+
alt_range = 100
|
327
|
+
alt_max = 0.5
|
328
|
+
dep_range = 100
|
329
|
+
dep_max = 500
|
330
|
+
|
331
|
+
|
332
|
+
for i in list
|
333
|
+
|
334
|
+
j = i[vcrid].to_f
|
335
|
+
d = i[depid].to_f
|
336
|
+
|
337
|
+
k = (j * alt_range / alt_max).to_i
|
338
|
+
alt_bin[k]=0 unless alt_bin[k]
|
339
|
+
alt_bin[k]+=1
|
340
|
+
|
341
|
+
end
|
342
|
+
|
343
|
+
|
344
|
+
dot = []
|
345
|
+
|
346
|
+
for i in list
|
347
|
+
|
348
|
+
j = i[vcrid].to_f
|
349
|
+
d = i[depid].to_f
|
350
|
+
k = (j * alt_range / alt_max).to_i
|
351
|
+
k2 = (d * dep_range / dep_max).to_i
|
352
|
+
|
353
|
+
dot << "[#{d},#{j}]"
|
354
|
+
|
355
|
+
|
356
|
+
end
|
357
|
+
|
358
|
+
|
359
|
+
|
360
|
+
|
361
|
+
return {:gc=>gc_bin,:dep=>d_bin,:alt=>alt_bin,:dot=>dot}
|
362
|
+
|
363
|
+
end
|
364
|
+
|
365
|
+
|
366
|
+
end
|
367
|
+
end
|
@@ -0,0 +1,122 @@
|
|
1
|
+
|
2
|
+
module Vardetect
|
3
|
+
module Vc
|
4
|
+
|
5
|
+
|
6
|
+
def self.call_prepare_st params
|
7
|
+
# ====================================================================================================
|
8
|
+
# 2. ST preparing filtered chromosome
|
9
|
+
# ====================================================================================================
|
10
|
+
count = 0
|
11
|
+
start = false
|
12
|
+
sam_tool_file =params['st']
|
13
|
+
snp_file =params['st_out']
|
14
|
+
|
15
|
+
|
16
|
+
st_output = File.open(snp_file,'w')
|
17
|
+
CSV.foreach(sam_tool_file) do |row|
|
18
|
+
if start
|
19
|
+
|
20
|
+
i = row[0][3..-1]
|
21
|
+
id = i.to_i
|
22
|
+
id = i if id == 0
|
23
|
+
key = "#{id}-#{row[1]}"
|
24
|
+
st_output.puts key+","+row.join(",")
|
25
|
+
|
26
|
+
count +=1
|
27
|
+
else
|
28
|
+
start=true
|
29
|
+
end
|
30
|
+
end
|
31
|
+
st_output.close
|
32
|
+
|
33
|
+
end
|
34
|
+
|
35
|
+
def self.call_prepare_snp params
|
36
|
+
# ====================================================================================================
|
37
|
+
# 3. SNP preparing filtered chromosome
|
38
|
+
# ====================================================================================================
|
39
|
+
|
40
|
+
snp_array_file =params['snp']
|
41
|
+
snp_file =params['snp_out']
|
42
|
+
|
43
|
+
count = 0
|
44
|
+
list = []
|
45
|
+
start = false
|
46
|
+
run = false
|
47
|
+
|
48
|
+
file = File.open(snp_array_file,'r')
|
49
|
+
output = File.open snp_file,'w'
|
50
|
+
map = {}
|
51
|
+
while str = file.gets #and count<100
|
52
|
+
str.strip!
|
53
|
+
if str=='[Data]' and !start
|
54
|
+
start = true
|
55
|
+
elsif start
|
56
|
+
s = str.split
|
57
|
+
if s[0] =='Sample' and !run
|
58
|
+
run = true
|
59
|
+
else
|
60
|
+
|
61
|
+
key = "#{s[1]}-#{s[2]}"
|
62
|
+
map[key] = s
|
63
|
+
|
64
|
+
end
|
65
|
+
|
66
|
+
end
|
67
|
+
count+=1
|
68
|
+
end
|
69
|
+
#
|
70
|
+
map.keys.sort.each do |k|
|
71
|
+
v = map[k]
|
72
|
+
output.puts "#{k},"+v[3..-1].join(",")
|
73
|
+
end
|
74
|
+
|
75
|
+
output.close
|
76
|
+
|
77
|
+
|
78
|
+
|
79
|
+
end
|
80
|
+
|
81
|
+
def self.call_prepare_vc params
|
82
|
+
|
83
|
+
vc_raw_file =params['vc_raw']
|
84
|
+
vc_file =params['vc_out']
|
85
|
+
|
86
|
+
count = 0
|
87
|
+
list = []
|
88
|
+
|
89
|
+
file = File.open(vc_raw_file,'r')
|
90
|
+
output = File.open vc_file,'w'
|
91
|
+
map = {}
|
92
|
+
while str = file.gets #and count<100
|
93
|
+
str.strip!
|
94
|
+
s = str.split
|
95
|
+
# puts str
|
96
|
+
begin
|
97
|
+
if str
|
98
|
+
|
99
|
+
alt = s[6].to_i
|
100
|
+
dep = s[5].to_i + alt
|
101
|
+
if ( alt.to_f/dep > 0.1) or s[2]!=s[3]
|
102
|
+
i = s[0][3..-1]
|
103
|
+
key = "#{i}-#{s[1]}"
|
104
|
+
map[key] = s
|
105
|
+
end
|
106
|
+
end
|
107
|
+
rescue
|
108
|
+
puts str
|
109
|
+
end
|
110
|
+
end
|
111
|
+
#
|
112
|
+
map.keys.sort.each do |k|
|
113
|
+
v = map[k]
|
114
|
+
output.puts "#{k},"+v[2..-1].join(",")
|
115
|
+
end
|
116
|
+
output.close
|
117
|
+
end
|
118
|
+
|
119
|
+
|
120
|
+
|
121
|
+
end
|
122
|
+
end
|
@@ -0,0 +1,493 @@
|
|
1
|
+
module Vardetect
|
2
|
+
module Vc
|
3
|
+
|
4
|
+
|
5
|
+
|
6
|
+
def self.form seq, cigar
|
7
|
+
ix = cigar.split(/S|M|I|D/)
|
8
|
+
index = 0
|
9
|
+
id = 0
|
10
|
+
str = ''
|
11
|
+
st = 0
|
12
|
+
indel = []
|
13
|
+
# puts seq
|
14
|
+
for i in ix
|
15
|
+
index+=i.size
|
16
|
+
cmd = cigar[index]
|
17
|
+
# puts "#{cmd} #{i.to_i}"
|
18
|
+
index+=1
|
19
|
+
i = i.to_i
|
20
|
+
case cmd
|
21
|
+
when 'S'
|
22
|
+
st = i
|
23
|
+
id += i
|
24
|
+
when 'I'
|
25
|
+
indel<<{:pos=>id-st,:type=>'ins',:size=>i,:cigar=>cigar}
|
26
|
+
id+=i
|
27
|
+
when 'D'
|
28
|
+
str+='N'*i
|
29
|
+
indel<<{:pos=>id-st,:type=>'del',:size=>i,:cigar=>cigar}
|
30
|
+
when 'M'
|
31
|
+
str += seq[id..id+i-1] if seq[id..id+i-1]
|
32
|
+
id+=i
|
33
|
+
end
|
34
|
+
# puts "Ref:#{seq}"
|
35
|
+
# puts "Seq:#{str}"
|
36
|
+
# puts "ID #{id}"
|
37
|
+
# puts
|
38
|
+
end
|
39
|
+
return str,indel
|
40
|
+
end
|
41
|
+
|
42
|
+
def self.call_sam_snv sam,chr,chr_start,chr_stop, debug=nil
|
43
|
+
|
44
|
+
unless debug
|
45
|
+
show = false
|
46
|
+
indel_show = false
|
47
|
+
var_show = false
|
48
|
+
else
|
49
|
+
s = debug.split('|')
|
50
|
+
show = s.index('show')
|
51
|
+
indel_show = s.index('indel')
|
52
|
+
var_show = s.index('var')
|
53
|
+
end
|
54
|
+
|
55
|
+
|
56
|
+
l = nil
|
57
|
+
snps = []
|
58
|
+
indels = []
|
59
|
+
indels_map = {}
|
60
|
+
|
61
|
+
puts "#{chr} #{chr_start} - #{chr_stop}"
|
62
|
+
seq = sam.fetch_reference(chr,chr_start,chr_stop+1000)
|
63
|
+
sam.load_index
|
64
|
+
|
65
|
+
upseq = seq.upcase
|
66
|
+
|
67
|
+
|
68
|
+
|
69
|
+
als = []
|
70
|
+
|
71
|
+
|
72
|
+
als = sam.fetch(chr, chr_start, chr_stop)
|
73
|
+
|
74
|
+
profile = []
|
75
|
+
list = []
|
76
|
+
start = nil
|
77
|
+
alleles = {'A'=>0,'C'=>1,'G'=>2,'T'=>3}
|
78
|
+
inv_alleles = {0=>'A',1=>'C',2=>'G',3=>'T'}
|
79
|
+
|
80
|
+
|
81
|
+
als.each{|i|
|
82
|
+
|
83
|
+
# puts "#{i.pos}"
|
84
|
+
|
85
|
+
unless start
|
86
|
+
start = i.pos
|
87
|
+
end
|
88
|
+
|
89
|
+
if i.pos!=start
|
90
|
+
|
91
|
+
while start<i.pos
|
92
|
+
del = []
|
93
|
+
|
94
|
+
s = ''
|
95
|
+
ref = seq[start-chr_start]
|
96
|
+
print "#{start}\t#{ref}\t#{list.size}\t" if show
|
97
|
+
ref = ref.upcase
|
98
|
+
|
99
|
+
|
100
|
+
p = [0,0,0,0]
|
101
|
+
q = [0,0,0,0]
|
102
|
+
|
103
|
+
list.each_with_index do|l,index|
|
104
|
+
|
105
|
+
|
106
|
+
c = l.seq[start-l.pos]
|
107
|
+
cq = l.qual[start-l.pos]
|
108
|
+
|
109
|
+
# a = seq[start-1..start-1-5]
|
110
|
+
# b =seq[start-l.pos..start-l.pos-5]
|
111
|
+
# puts "#{a} - #{b}"
|
112
|
+
|
113
|
+
if c
|
114
|
+
|
115
|
+
case c
|
116
|
+
when 'A'
|
117
|
+
p[0]+=1
|
118
|
+
q[0]+=cq-33 if cq
|
119
|
+
when 'C'
|
120
|
+
p[1]+=1
|
121
|
+
q[1]+=cq-33 if cq
|
122
|
+
when 'G'
|
123
|
+
p[2]+=1
|
124
|
+
q[2]+=cq-33 if cq
|
125
|
+
when 'T'
|
126
|
+
p[3]+=1
|
127
|
+
q[3]+=cq-33 if cq
|
128
|
+
end
|
129
|
+
|
130
|
+
s+=c
|
131
|
+
|
132
|
+
print "#{c}" if show
|
133
|
+
|
134
|
+
else
|
135
|
+
del<<l
|
136
|
+
end
|
137
|
+
|
138
|
+
|
139
|
+
end
|
140
|
+
|
141
|
+
|
142
|
+
|
143
|
+
pmax = p.index(p.max)
|
144
|
+
max = p[pmax]
|
145
|
+
p[pmax] = 0 if pmax
|
146
|
+
palt = p.index(p.max)
|
147
|
+
alt = p[palt]
|
148
|
+
dep = max+alt
|
149
|
+
max_q = q[pmax].to_f/max
|
150
|
+
alt_q = '0'
|
151
|
+
alt_q = q[palt].to_f/alt if alt > 0
|
152
|
+
|
153
|
+
|
154
|
+
if dep > 2 and ((indel = indels_map[start] and indel[:count]>1 ) or (alt>0 and alt/dep.to_f > 0.1 and alt_q >10) or alleles[ref]!=pmax )
|
155
|
+
|
156
|
+
|
157
|
+
|
158
|
+
text = "#{start}\t#{max}/#{alt.to_i}\t#{ref}\t#{s}"
|
159
|
+
puts text if var_show
|
160
|
+
sum = (max+alt).to_f
|
161
|
+
aalt = '-'
|
162
|
+
|
163
|
+
aalt = inv_alleles[palt] if alt!=0
|
164
|
+
if (alt>0 and alt/dep.to_f > 0.1)
|
165
|
+
c = 'a'
|
166
|
+
else
|
167
|
+
c = 'A'
|
168
|
+
end
|
169
|
+
|
170
|
+
|
171
|
+
if indel
|
172
|
+
s = indel[:alleles].split('/')
|
173
|
+
ref = s[0]
|
174
|
+
aalt = s[1]
|
175
|
+
alt = indel[:count]
|
176
|
+
if indel[:type]=='ins'
|
177
|
+
c='+'
|
178
|
+
else
|
179
|
+
c='-'
|
180
|
+
end
|
181
|
+
end
|
182
|
+
snps << [chr,start,ref,inv_alleles[pmax],aalt,max,alt,format('%.2f',max_q),format('%.2f',alt_q),c]
|
183
|
+
|
184
|
+
# snps << [chr,start,ref,inv_alleles[pmax],inv_alleles[palt],max,alt,max/sum,alt/sum]
|
185
|
+
|
186
|
+
end
|
187
|
+
list-=del
|
188
|
+
if list.size==0
|
189
|
+
start = i.pos
|
190
|
+
else
|
191
|
+
start+=1
|
192
|
+
end
|
193
|
+
|
194
|
+
|
195
|
+
puts if show
|
196
|
+
end
|
197
|
+
|
198
|
+
start = i.pos
|
199
|
+
end
|
200
|
+
|
201
|
+
|
202
|
+
|
203
|
+
seq_size = i.seq.size
|
204
|
+
|
205
|
+
tseq = form(i.seq,i.cigar)
|
206
|
+
tqual = form(i.qual,i.cigar)
|
207
|
+
|
208
|
+
|
209
|
+
if (i.cigar.index('I') or i.cigar.index('D') )
|
210
|
+
|
211
|
+
|
212
|
+
if indel_show
|
213
|
+
puts "-------------------------INDEL "
|
214
|
+
puts i.pos
|
215
|
+
puts i.cigar
|
216
|
+
puts i.seq
|
217
|
+
puts i.qual
|
218
|
+
# puts i.inspect
|
219
|
+
|
220
|
+
# :qname, :flag, :rname,:pos,:mapq,:cigar, :mrnm, :mpos, :isize, :seq, :qual, :tags, :al, :samstr
|
221
|
+
|
222
|
+
puts seq[i.pos-chr_start..i.pos-chr_start+i.seq.size]
|
223
|
+
puts tseq[0]
|
224
|
+
puts tqual[0]
|
225
|
+
puts tseq[1]
|
226
|
+
end
|
227
|
+
|
228
|
+
if tseq[1].size>0
|
229
|
+
for y in tseq[1]
|
230
|
+
if y[:type]=='ins'
|
231
|
+
iref = '-'
|
232
|
+
iseq = i.seq[y[:pos]..y[:pos]+y[:size]-1]
|
233
|
+
y[:alleles] = "#{iref}/#{iseq}"
|
234
|
+
else
|
235
|
+
iref = seq[i.pos-chr_start+y[:pos]..i.pos-chr_start+y[:pos]+y[:size]-1]
|
236
|
+
iseq = '-'
|
237
|
+
y[:alleles] = "#{iref}/#{iseq}"
|
238
|
+
end
|
239
|
+
y[:pos]+=i.pos
|
240
|
+
indels<<y
|
241
|
+
unless indels_map[y[:pos]]
|
242
|
+
y[:count]=1
|
243
|
+
y[:sm]=y[:alleles]
|
244
|
+
indels_map[y[:pos]] = y
|
245
|
+
else
|
246
|
+
indels_map[y[:pos]][:sm]+=",#{y[:alleles]}"
|
247
|
+
|
248
|
+
indels_map[y[:pos]][:count] +=1
|
249
|
+
end
|
250
|
+
|
251
|
+
end
|
252
|
+
end
|
253
|
+
|
254
|
+
|
255
|
+
end
|
256
|
+
|
257
|
+
|
258
|
+
|
259
|
+
|
260
|
+
|
261
|
+
i.seq = tseq[0]
|
262
|
+
i.qual = tqual[0].bytes.to_a
|
263
|
+
|
264
|
+
|
265
|
+
if i.seq.size > seq_size/2
|
266
|
+
|
267
|
+
si = 0
|
268
|
+
|
269
|
+
i.seq.size.times do |k|
|
270
|
+
# puts "#{i.seq[k]} #{seq[start-chr_start+k]}"
|
271
|
+
si+=1 if i.seq[k]==upseq[start-chr_start+k]
|
272
|
+
end
|
273
|
+
# puts "#{si} #{i.seq.size}"
|
274
|
+
if (si.to_f/i.seq.size)>=0.95
|
275
|
+
list << i
|
276
|
+
else
|
277
|
+
if indel_show
|
278
|
+
puts "Drop with low similarity #{si.to_f/i.seq.size} #{i.cigar}"
|
279
|
+
puts seq[start-chr_start..start-chr_start+100]
|
280
|
+
puts i.seq
|
281
|
+
end
|
282
|
+
end
|
283
|
+
end
|
284
|
+
|
285
|
+
l = i
|
286
|
+
|
287
|
+
|
288
|
+
}
|
289
|
+
|
290
|
+
|
291
|
+
if indel_show
|
292
|
+
|
293
|
+
for i in indels_map.keys.sort
|
294
|
+
puts indels_map[i].inspect
|
295
|
+
end
|
296
|
+
end
|
297
|
+
# puts l.cigar
|
298
|
+
# puts l.seq
|
299
|
+
# puts snps.size
|
300
|
+
return snps
|
301
|
+
|
302
|
+
end
|
303
|
+
|
304
|
+
def self.call_variance p
|
305
|
+
|
306
|
+
|
307
|
+
chr = p[:chr]
|
308
|
+
|
309
|
+
output = p[:output]
|
310
|
+
testBAMFile = p[:sam]
|
311
|
+
testReference = p[:ref]
|
312
|
+
sam = Bio::DB::Sam.new({:bam=>testBAMFile, :fasta=>testReference})
|
313
|
+
sam.open
|
314
|
+
|
315
|
+
chr_start = p[:start]
|
316
|
+
chr_stop = p[:stop]
|
317
|
+
snps = call_sam_snv sam,chr,chr_start,chr_stop, p[:debug]
|
318
|
+
|
319
|
+
sam.close
|
320
|
+
f = File.open(output,'a')
|
321
|
+
f.puts snps.collect{|j| j.join("\t")}
|
322
|
+
f.close
|
323
|
+
end
|
324
|
+
|
325
|
+
|
326
|
+
|
327
|
+
def self.inspect_reference testReference
|
328
|
+
|
329
|
+
# index reference
|
330
|
+
|
331
|
+
file = File.open(testReference,'r')
|
332
|
+
map = {}
|
333
|
+
size=0
|
334
|
+
chr = nil
|
335
|
+
while str=file.gets
|
336
|
+
if str[0..0]=='>'
|
337
|
+
if chr
|
338
|
+
puts "#{chr}\t#{size}"
|
339
|
+
map[chr]=size
|
340
|
+
end
|
341
|
+
chr = str.strip.split[0][1..-1]
|
342
|
+
size=0
|
343
|
+
else
|
344
|
+
size+=str.strip.size
|
345
|
+
end
|
346
|
+
end
|
347
|
+
map[chr]=size
|
348
|
+
puts map.inspect
|
349
|
+
file.close
|
350
|
+
genome = map
|
351
|
+
# genome = {"chr1"=>249250621, "chr10"=>135534747, "chr11"=>135006516, "chr12"=>133851895, "chr13"=>115169878, "chr14"=>107349540, "chr15"=>102531392, "chr16"=>90354753, "chr17"=>81195210, "chr18"=>78077248, "chr19"=>59128983, "chr2"=>243199373, "chr20"=>63025520, "chr21"=>48129895, "chr22"=>51304566, "chr3"=>198022430, "chr4"=>191154276, "chr5"=>180915260, "chr6"=>171115067, "chr7"=>159138663, "chr8"=>146364022, "chr9"=>141213431, "chrX"=>155270560, "chrY"=>59373566}
|
352
|
+
|
353
|
+
return genome
|
354
|
+
end
|
355
|
+
|
356
|
+
def self.inspect_reference_human testReference
|
357
|
+
|
358
|
+
# index reference
|
359
|
+
#
|
360
|
+
# file = File.open(testReference,'r')
|
361
|
+
# map = {}
|
362
|
+
# size=0
|
363
|
+
# chr = nil
|
364
|
+
# while str=file.gets
|
365
|
+
# if str[0..0]=='>'
|
366
|
+
# if chr
|
367
|
+
# puts "#{chr}\t#{size}"
|
368
|
+
# map[chr]=size
|
369
|
+
# end
|
370
|
+
# chr = str[1..-1].strip
|
371
|
+
# size=0
|
372
|
+
# else
|
373
|
+
# size+=str.strip.size
|
374
|
+
# end
|
375
|
+
# end
|
376
|
+
# map[chr]=size
|
377
|
+
# puts map.inspect
|
378
|
+
# file.close
|
379
|
+
|
380
|
+
genome = {"chr1"=>249250621, "chr10"=>135534747, "chr11"=>135006516, "chr12"=>133851895, "chr13"=>115169878, "chr14"=>107349540, "chr15"=>102531392, "chr16"=>90354753, "chr17"=>81195210, "chr18"=>78077248, "chr19"=>59128983, "chr2"=>243199373, "chr20"=>63025520, "chr21"=>48129895, "chr22"=>51304566, "chr3"=>198022430, "chr4"=>191154276, "chr5"=>180915260, "chr6"=>171115067, "chr7"=>159138663, "chr8"=>146364022, "chr9"=>141213431, "chrX"=>155270560, "chrY"=>59373566}
|
381
|
+
|
382
|
+
return genome
|
383
|
+
end
|
384
|
+
|
385
|
+
|
386
|
+
def self.process_vc_multicore params, core=1,chr_filter=nil
|
387
|
+
|
388
|
+
sim = core
|
389
|
+
|
390
|
+
path = '/Users/soup/Desktop/vcqa'
|
391
|
+
output = params[:output]
|
392
|
+
|
393
|
+
testReference = params[:ref]
|
394
|
+
|
395
|
+
genome = inspect_reference_human(testReference)
|
396
|
+
|
397
|
+
if chr_filter # 1 as text
|
398
|
+
|
399
|
+
tmp = {}
|
400
|
+
chr = "chr#{chr_filter}"
|
401
|
+
tmp[chr]=genome[chr]
|
402
|
+
genome = tmp
|
403
|
+
end
|
404
|
+
|
405
|
+
puts genome.inspect
|
406
|
+
|
407
|
+
|
408
|
+
# reset output
|
409
|
+
f = File.open(output,'w')
|
410
|
+
f.close
|
411
|
+
|
412
|
+
a = []
|
413
|
+
|
414
|
+
# job generator
|
415
|
+
genome.each_pair do |chr,size|
|
416
|
+
puts "#{chr}\t#{size}"
|
417
|
+
m = 1000000
|
418
|
+
n = size/m
|
419
|
+
|
420
|
+
n.times do |i|
|
421
|
+
params[:chr] = chr
|
422
|
+
params[:start] = 1 + i * m
|
423
|
+
params[:stop] = (i+1) * m
|
424
|
+
a<<params.clone
|
425
|
+
end
|
426
|
+
|
427
|
+
if size%m !=0
|
428
|
+
params[:chr] = chr
|
429
|
+
params[:start] = size - size% m
|
430
|
+
params[:stop] = size
|
431
|
+
a<<params.clone
|
432
|
+
end
|
433
|
+
|
434
|
+
end
|
435
|
+
|
436
|
+
sim = a.size if sim>a.size
|
437
|
+
|
438
|
+
# starting first N processes
|
439
|
+
sim.times do
|
440
|
+
i = a.pop
|
441
|
+
Process.fork{call_variance(i)}
|
442
|
+
end
|
443
|
+
# start one by one as the previous finish
|
444
|
+
a.each do |i|
|
445
|
+
Process.wait(0)
|
446
|
+
Process.fork{call_variance(i)}
|
447
|
+
end
|
448
|
+
# wait for all to finish
|
449
|
+
Process.waitall
|
450
|
+
|
451
|
+
end
|
452
|
+
|
453
|
+
|
454
|
+
|
455
|
+
def self.call_snv params
|
456
|
+
# ====================================================================================================
|
457
|
+
# 1. VC calling
|
458
|
+
# ====================================================================================================
|
459
|
+
core = 4
|
460
|
+
core = params['core'].to_i if params['core'] # number of simultaneous processes
|
461
|
+
sam_file =params['sam']
|
462
|
+
ref_file =params['ref']
|
463
|
+
vc_file =params['vc_out']
|
464
|
+
p = {:sam=>sam_file,:ref=>ref_file,:output=>vc_file}
|
465
|
+
|
466
|
+
# params[:start] = 95_658_000
|
467
|
+
# params[:stop] = 95_659_000
|
468
|
+
# call_variance params
|
469
|
+
|
470
|
+
process_vc_multicore p,core
|
471
|
+
# # params = {:sam=>sam_file,:ref=>ref_file,:output=>vc_all_file}
|
472
|
+
# # process_vc_multicore params,core
|
473
|
+
end
|
474
|
+
|
475
|
+
|
476
|
+
def self.call_snv_range params
|
477
|
+
# ====================================================================================================
|
478
|
+
# 1. VC calling
|
479
|
+
# ====================================================================================================
|
480
|
+
core = 4
|
481
|
+
core = params['core'].to_i if params['core'] # number of simultaneous processes
|
482
|
+
sam_file =params['sam']
|
483
|
+
ref_file =params['ref']
|
484
|
+
vc_file =params['vc_out']
|
485
|
+
p = {:sam=>sam_file,:ref=>ref_file,:output=>vc_file,:chr=>"chr#{params['chr']}",:start=>params['start'].to_i,:stop=>params['stop'].to_i,:debug=>'indel'}
|
486
|
+
puts p.inspect
|
487
|
+
|
488
|
+
call_variance p
|
489
|
+
|
490
|
+
end
|
491
|
+
|
492
|
+
end
|
493
|
+
end
|
@@ -0,0 +1,28 @@
|
|
1
|
+
require 'vardetect-vc/call_snv'
|
2
|
+
require 'vardetect-vc/call_prepare'
|
3
|
+
require 'vardetect-vc/call_combine'
|
4
|
+
|
5
|
+
module Vardetect
|
6
|
+
module Vc
|
7
|
+
|
8
|
+
|
9
|
+
def self.parse_params params
|
10
|
+
|
11
|
+
puts params
|
12
|
+
params = params.split
|
13
|
+
cmd = params[0]
|
14
|
+
tmp = {}
|
15
|
+
((params.size-1)/2).times do |i|
|
16
|
+
key = params[1+i*2][1..-1]
|
17
|
+
value = params[1+i*2+1]
|
18
|
+
tmp[key]=value
|
19
|
+
end
|
20
|
+
tmp[:cmd]=cmd
|
21
|
+
params = tmp
|
22
|
+
end
|
23
|
+
|
24
|
+
end
|
25
|
+
|
26
|
+
end
|
27
|
+
|
28
|
+
|
data/lib/vardetect-vc/version.rb
CHANGED
metadata
CHANGED
@@ -1,7 +1,7 @@
|
|
1
1
|
--- !ruby/object:Gem::Specification
|
2
2
|
name: vardetect-vc
|
3
3
|
version: !ruby/object:Gem::Version
|
4
|
-
version: 0.0.
|
4
|
+
version: 0.0.5
|
5
5
|
prerelease:
|
6
6
|
platform: ruby
|
7
7
|
authors:
|
@@ -78,7 +78,9 @@ dependencies:
|
|
78
78
|
description: A variance calling tool for NGS
|
79
79
|
email:
|
80
80
|
- supasak.kul@biotec.or.th
|
81
|
-
executables:
|
81
|
+
executables:
|
82
|
+
- vardetect_vc
|
83
|
+
- vardetect_vc_batch
|
82
84
|
extensions: []
|
83
85
|
extra_rdoc_files: []
|
84
86
|
files:
|
@@ -87,7 +89,13 @@ files:
|
|
87
89
|
- LICENSE.txt
|
88
90
|
- README.md
|
89
91
|
- Rakefile
|
92
|
+
- bin/vardetect_vc
|
93
|
+
- bin/vardetect_vc_batch
|
90
94
|
- lib/vardetect-vc.rb
|
95
|
+
- lib/vardetect-vc/call_combine.rb
|
96
|
+
- lib/vardetect-vc/call_prepare.rb
|
97
|
+
- lib/vardetect-vc/call_snv.rb
|
98
|
+
- lib/vardetect-vc/lib.rb
|
91
99
|
- lib/vardetect-vc/version.rb
|
92
100
|
- vardetect-vc.gemspec
|
93
101
|
homepage: ''
|