gene_assembler 0.0.9
Sign up to get free protection for your applications and to get access to all the features.
- data/.gitignore +22 -0
- data/Gemfile +4 -0
- data/LICENSE.txt +22 -0
- data/README.md +29 -0
- data/Rakefile +2 -0
- data/bin/GeneAssembler +233 -0
- data/bin/phytozome_scan +60 -0
- data/gene_assembler.gemspec +25 -0
- data/lib/gene_assembler.rb +5 -0
- data/lib/gene_assembler/blast_type_parser.rb +41 -0
- data/lib/gene_assembler/contig.rb +643 -0
- data/lib/gene_assembler/dataset.rb +532 -0
- data/lib/gene_assembler/exonerate_result.rb +230 -0
- data/lib/gene_assembler/gff_contig.rb +67 -0
- data/lib/gene_assembler/gff_dataset.rb +152 -0
- data/lib/gene_assembler/gff_feature.rb +175 -0
- data/lib/gene_assembler/gff_frameshift.rb +6 -0
- data/lib/gene_assembler/gff_go.rb +13 -0
- data/lib/gene_assembler/gff_hit.rb +53 -0
- data/lib/gene_assembler/gff_hsp.rb +6 -0
- data/lib/gene_assembler/gff_localization.rb +6 -0
- data/lib/gene_assembler/gff_master_feature.rb +5 -0
- data/lib/gene_assembler/gff_parser.rb +35 -0
- data/lib/gene_assembler/gff_snp.rb +21 -0
- data/lib/gene_assembler/gff_stop.rb +6 -0
- data/lib/gene_assembler/go.rb +13 -0
- data/lib/gene_assembler/hit.rb +191 -0
- data/lib/gene_assembler/hsp.rb +100 -0
- data/lib/gene_assembler/other_functions.rb +228 -0
- data/lib/gene_assembler/parser.rb +25 -0
- data/lib/gene_assembler/parser_blast.rb +12 -0
- data/lib/gene_assembler/parser_exonerate.rb +16 -0
- data/lib/gene_assembler/rebuild.rb +975 -0
- data/lib/gene_assembler/report.rb +13 -0
- data/lib/gene_assembler/report_gff.rb +30 -0
- data/lib/gene_assembler/snp.rb +13 -0
- data/lib/gene_assembler/version.rb +3 -0
- metadata +149 -0
@@ -0,0 +1,643 @@
|
|
1
|
+
require 'hit'
|
2
|
+
require 'snp'
|
3
|
+
require 'go'
|
4
|
+
|
5
|
+
|
6
|
+
class Contig
|
7
|
+
attr_accessor :name, :seq,:type, :length, :hits, :completed, :q_frameshift, :s_frameshift, :stops, :mod_coord
|
8
|
+
def initialize (name)
|
9
|
+
@name=name
|
10
|
+
@seq=''
|
11
|
+
@type=nil
|
12
|
+
@length=''
|
13
|
+
@hits=[]
|
14
|
+
@snps=[]
|
15
|
+
@gos=[]
|
16
|
+
@completed=''
|
17
|
+
@localization=[]
|
18
|
+
@q_frameshift=[]
|
19
|
+
@s_frameshift=[]
|
20
|
+
@stops=[]
|
21
|
+
@mod_coord=FALSE #Indica si se han alterado las coordenadas del contig previamente
|
22
|
+
end
|
23
|
+
|
24
|
+
def length=(length)
|
25
|
+
if length.class.to_s=='String'
|
26
|
+
@length=length.to_i
|
27
|
+
else
|
28
|
+
@length=length
|
29
|
+
end
|
30
|
+
end
|
31
|
+
|
32
|
+
def first_hit
|
33
|
+
h=nil
|
34
|
+
each_hit{|hit|
|
35
|
+
h=hit
|
36
|
+
break
|
37
|
+
}
|
38
|
+
return h
|
39
|
+
end
|
40
|
+
|
41
|
+
def last_hit
|
42
|
+
h=nil
|
43
|
+
each_hit{|hit|
|
44
|
+
h=hit
|
45
|
+
}
|
46
|
+
return h
|
47
|
+
end
|
48
|
+
|
49
|
+
def hit_count
|
50
|
+
count=0
|
51
|
+
each_hit{|hit|
|
52
|
+
count+=1
|
53
|
+
}
|
54
|
+
return count
|
55
|
+
end
|
56
|
+
|
57
|
+
def add_localization(localization)
|
58
|
+
@localization << localization
|
59
|
+
end
|
60
|
+
|
61
|
+
def each_localization
|
62
|
+
@localization.each do |localization|
|
63
|
+
yield localization
|
64
|
+
end
|
65
|
+
end
|
66
|
+
|
67
|
+
def each_localization_with_index
|
68
|
+
@localization.each_with_index do |localization,i|
|
69
|
+
yield localization,i
|
70
|
+
end
|
71
|
+
end
|
72
|
+
|
73
|
+
def each_stop
|
74
|
+
@stops.each do |stop|
|
75
|
+
yield stop
|
76
|
+
end
|
77
|
+
end
|
78
|
+
|
79
|
+
def each_stop_with_index
|
80
|
+
@stops.each_with_index do |stop,i|
|
81
|
+
yield stop,i
|
82
|
+
end
|
83
|
+
end
|
84
|
+
|
85
|
+
def add_go(go,name,obsolete)
|
86
|
+
go=GO.new(go,name,obsolete)
|
87
|
+
@gos << go
|
88
|
+
return go
|
89
|
+
end
|
90
|
+
|
91
|
+
def each_go
|
92
|
+
@gos.each do |go|
|
93
|
+
yield go
|
94
|
+
end
|
95
|
+
end
|
96
|
+
|
97
|
+
def add_snp(position)
|
98
|
+
snp=SNP.new(position)
|
99
|
+
@snps << snp
|
100
|
+
return snp
|
101
|
+
end
|
102
|
+
|
103
|
+
def add_hit(hit_name, s_length, reversed, type)
|
104
|
+
hit=Hit.new(hit_name, s_length, reversed, type)
|
105
|
+
@hits << hit
|
106
|
+
return hit
|
107
|
+
end
|
108
|
+
|
109
|
+
def has_hit?
|
110
|
+
has_hit=FALSE
|
111
|
+
if @hits.length>0
|
112
|
+
has_hit=TRUE
|
113
|
+
end
|
114
|
+
return has_hit
|
115
|
+
end
|
116
|
+
|
117
|
+
def each_hit
|
118
|
+
@hits.each do |hit|
|
119
|
+
yield hit
|
120
|
+
end
|
121
|
+
end
|
122
|
+
|
123
|
+
def each_hit_with_index
|
124
|
+
@hits.each_with_index do |hit,i|
|
125
|
+
yield hit,i
|
126
|
+
end
|
127
|
+
end
|
128
|
+
|
129
|
+
def n_hits?
|
130
|
+
n=@hits.length
|
131
|
+
return n
|
132
|
+
end
|
133
|
+
|
134
|
+
def each_snp
|
135
|
+
@snps.each do |snp|
|
136
|
+
yield snp
|
137
|
+
end
|
138
|
+
end
|
139
|
+
|
140
|
+
def each_snp_with_index
|
141
|
+
@snps.each_with_index do |snp,i|
|
142
|
+
yield snp,i
|
143
|
+
end
|
144
|
+
end
|
145
|
+
|
146
|
+
def hits_sort!
|
147
|
+
each_hit{|hit|
|
148
|
+
hit.hsps_sort!
|
149
|
+
}
|
150
|
+
end
|
151
|
+
|
152
|
+
def modified_coordenates(add)
|
153
|
+
@mod_coord=TRUE
|
154
|
+
each_hit{|hit|
|
155
|
+
hit.modified_coordenates(add)
|
156
|
+
stop_modified_coordenates(add)
|
157
|
+
frameshift_modified_coordenates(add)
|
158
|
+
}
|
159
|
+
return last_hit.last_hsp.q_end
|
160
|
+
end
|
161
|
+
|
162
|
+
def stop_modified_coordenates(add)
|
163
|
+
@stops.length.times do |n|
|
164
|
+
@stops[n]+=add
|
165
|
+
end
|
166
|
+
end
|
167
|
+
|
168
|
+
def frameshift_modified_coordenates(add)
|
169
|
+
@q_frameshift.length.times do |n|
|
170
|
+
@q_frameshift[n]+=add
|
171
|
+
end
|
172
|
+
end
|
173
|
+
|
174
|
+
def each_q_frameshift
|
175
|
+
@q_frameshift.each do |qfs|
|
176
|
+
yield qfs
|
177
|
+
end
|
178
|
+
end
|
179
|
+
|
180
|
+
def draw #Realiza una representacion del contig a nivel del subject
|
181
|
+
last_hsp_end=0
|
182
|
+
overlap=0
|
183
|
+
first_hit.each_hsp_with_index{|hsp,c|
|
184
|
+
dif=hsp.s_beg-last_hsp_end
|
185
|
+
if dif>=0 && dif<=2
|
186
|
+
print '/' # Limite solapante uno a continuacion de otro (disposicion normal del exon)
|
187
|
+
elsif dif>2
|
188
|
+
print '-'*(hsp.s_beg-last_hsp_end)
|
189
|
+
print '|'
|
190
|
+
elsif dif<0
|
191
|
+
print '&'
|
192
|
+
overlap=dif
|
193
|
+
end
|
194
|
+
h=c+1 #Num de hsp
|
195
|
+
print "#{h.to_s.center(hsp.s_end-hsp.s_beg+overlap-1)}"
|
196
|
+
if dif>2
|
197
|
+
#print '|'
|
198
|
+
end
|
199
|
+
last_hsp_end=hsp.s_end
|
200
|
+
}
|
201
|
+
print '|-'
|
202
|
+
ending=first_hit.s_length-last_hsp_end
|
203
|
+
if ending<0
|
204
|
+
ending=0
|
205
|
+
end
|
206
|
+
print '-'*(ending)
|
207
|
+
puts "\n"
|
208
|
+
end
|
209
|
+
|
210
|
+
def indices #Muestra los indices de subject y query del contig
|
211
|
+
each_hit_with_index {|hit,ind|
|
212
|
+
hit.each_hsp_with_index{|hsp,i|
|
213
|
+
puts "#{ind+1}.#{i+1})\t#{hsp.q_beg}\t#{hsp.q_end}\t#{hsp.s_beg}\t#{hsp.s_end}\t#{@name}\t#{@length}\t#{@mod_coord}"
|
214
|
+
}
|
215
|
+
}
|
216
|
+
end
|
217
|
+
|
218
|
+
def exones_s # Devuelve un array con el tamaño de cada hsp/exon a nivel del subject
|
219
|
+
exones_s=[]
|
220
|
+
each_hit {|hit|
|
221
|
+
hit.each_hsp{|hsp|
|
222
|
+
long=(hsp.s_end-hsp.s_beg).abs
|
223
|
+
exones_s << long
|
224
|
+
}
|
225
|
+
}
|
226
|
+
return exones_s
|
227
|
+
end
|
228
|
+
|
229
|
+
def exones_q # Devuelve un array con el tamaño de cada hsp/exon a nivel del query
|
230
|
+
exones_q=[]
|
231
|
+
each_hit {|hit|
|
232
|
+
hit.each_hsp{|hsp|
|
233
|
+
long=(hsp.q_end-hsp.q_beg).abs
|
234
|
+
exones_q << long
|
235
|
+
}
|
236
|
+
}
|
237
|
+
return exones_q
|
238
|
+
end
|
239
|
+
|
240
|
+
def intrones_q # Devuelve un array con el tamaño de cada intron a nivel del query
|
241
|
+
intrones_q=[]
|
242
|
+
l=first_hit.hsp_count
|
243
|
+
each_hit {|hit|
|
244
|
+
hit.each_hsp_with_index{|hsp,ind|
|
245
|
+
if !first_hit.hsp_at(ind+1)
|
246
|
+
break
|
247
|
+
end
|
248
|
+
long=(first_hit.hsp_at(ind+1).q_beg-hsp.q_end).abs
|
249
|
+
intrones_q << long
|
250
|
+
}
|
251
|
+
}
|
252
|
+
return intrones_q
|
253
|
+
end
|
254
|
+
|
255
|
+
def n_intron
|
256
|
+
n_intron=0
|
257
|
+
each_hit{|hit|
|
258
|
+
n_intron+=1
|
259
|
+
}
|
260
|
+
return n_intron
|
261
|
+
end
|
262
|
+
|
263
|
+
def exon_acumulative #Suma la longitud de todos los exones
|
264
|
+
long=0
|
265
|
+
exones=exones_q
|
266
|
+
exones.each do |ex|
|
267
|
+
long+=ex
|
268
|
+
end
|
269
|
+
return long
|
270
|
+
end
|
271
|
+
|
272
|
+
#Funciones para comprobar validez de los contig
|
273
|
+
def mixed? #Examina si los hsp de un hit estan desordenados
|
274
|
+
is_mix=FALSE
|
275
|
+
beg=nil
|
276
|
+
sign=0
|
277
|
+
sign_local=0
|
278
|
+
@hits.first.hsps.each_with_index do |hsp,c|
|
279
|
+
if !beg.nil?
|
280
|
+
dif=hsp.q_beg-beg
|
281
|
+
if dif>=0
|
282
|
+
sign_local=1
|
283
|
+
else sign_local=0
|
284
|
+
end
|
285
|
+
if sign_local!=sign && c>1
|
286
|
+
is_mix=TRUE
|
287
|
+
break
|
288
|
+
end
|
289
|
+
end
|
290
|
+
sign=sign_local
|
291
|
+
beg=hsp.q_beg
|
292
|
+
end
|
293
|
+
return is_mix
|
294
|
+
end
|
295
|
+
|
296
|
+
def is_one_hsp? #Examina si el hit esta compuesto por un solo hsp
|
297
|
+
is_one=FALSE
|
298
|
+
if first_hit.hsp_count==1
|
299
|
+
is_one=TRUE
|
300
|
+
end
|
301
|
+
return is_one
|
302
|
+
end
|
303
|
+
|
304
|
+
def is_gapped? #Examina si hay gaps internos en la estructura del gen q se mapea sobre la proteina q pudieran señalar la falta parcial o completa de un exon
|
305
|
+
gap=3 #Gap maximo permitido medido en aa, como minimo poner a 1
|
306
|
+
gapped=FALSE
|
307
|
+
s_end_last=0
|
308
|
+
@hits.first.hsps.each do |hsp|
|
309
|
+
if s_end_last >0
|
310
|
+
dif=hsp.s_beg-s_end_last
|
311
|
+
if dif>gap #En caso de coordenadas solapantes, siempre saldra dif negativo, con lo que la condicion gap no se cumple
|
312
|
+
gapped=TRUE
|
313
|
+
break
|
314
|
+
end
|
315
|
+
end
|
316
|
+
s_end_last=hsp.s_end
|
317
|
+
end
|
318
|
+
return gapped
|
319
|
+
end
|
320
|
+
|
321
|
+
def is_truncated? #Examina si los exones en el borde del contig estan truncados/interrumpidos
|
322
|
+
truncated=FALSE
|
323
|
+
# puts self.name
|
324
|
+
# puts @hits.inspect
|
325
|
+
#Truncado por el inicio
|
326
|
+
if first_hit.first_hsp.s_beg>1 && first_hit.first_hsp.q_beg==1
|
327
|
+
truncated=TRUE
|
328
|
+
end
|
329
|
+
|
330
|
+
#Truncado por el final
|
331
|
+
if first_hit.last_hsp.q_end==@length
|
332
|
+
truncated=TRUE
|
333
|
+
end
|
334
|
+
return truncated
|
335
|
+
end
|
336
|
+
|
337
|
+
def hsp_minor_than?(hsp_length) # En nt
|
338
|
+
minor=FALSE
|
339
|
+
each_hit_with_index {|hit,i|
|
340
|
+
if i>0 || i<hit.hsp_count
|
341
|
+
if hit.hsp_minor_than?(hsp_length)
|
342
|
+
minor=TRUE
|
343
|
+
break
|
344
|
+
end
|
345
|
+
end
|
346
|
+
}
|
347
|
+
return minor
|
348
|
+
end
|
349
|
+
|
350
|
+
def correct_hsps(blast_coor_type)
|
351
|
+
each_hit {|hit|
|
352
|
+
hit.correct_hsps(blast_coor_type)
|
353
|
+
}
|
354
|
+
end
|
355
|
+
#-------------------------------------------------------------------------
|
356
|
+
|
357
|
+
def compare(contig) #Alinea un contig con otro en base a las coordenadas del subject
|
358
|
+
exon_match=-1
|
359
|
+
exones=0
|
360
|
+
match_found=FALSE
|
361
|
+
#SELF HIT
|
362
|
+
self.each_hit {|self_hit|
|
363
|
+
if match_found #Romper bucle si ha habido coindidencia definitiva
|
364
|
+
break
|
365
|
+
end
|
366
|
+
#SELF HSP
|
367
|
+
self_hit.each_hsp {|self_hsp|
|
368
|
+
if match_found
|
369
|
+
break
|
370
|
+
end
|
371
|
+
#CONTIG HIT
|
372
|
+
last=0
|
373
|
+
contig_hsp_count=0
|
374
|
+
contig.each_hit {|contig_hit|
|
375
|
+
#CONTIG HSP
|
376
|
+
contig_hit.each_hsp {|contig_hsp|#Valoracion del coverage de cada hit entre si, en el momento que el segundo de mayor se cancela
|
377
|
+
coverage=self_hsp.compare(contig_hsp)
|
378
|
+
if coverage>last #Guardar coincidencia a la espera de una mejor
|
379
|
+
exon_match=contig_hsp_count
|
380
|
+
last=coverage
|
381
|
+
end
|
382
|
+
if coverage>0.2 #Contaje de exones
|
383
|
+
exones+=1
|
384
|
+
end
|
385
|
+
if coverage==0 && exon_match>-1 #Romper bucle cuando se ha encontrado coincidencia y los demas exones no coinciden
|
386
|
+
match_found=TRUE
|
387
|
+
break
|
388
|
+
end
|
389
|
+
contig_hsp_count+=1
|
390
|
+
}
|
391
|
+
if match_found
|
392
|
+
break
|
393
|
+
end
|
394
|
+
}
|
395
|
+
}
|
396
|
+
}
|
397
|
+
return exon_match,exones
|
398
|
+
end
|
399
|
+
|
400
|
+
def rev_comp_if_hit #Devuelve la secuencia reversocomplementaria del contig
|
401
|
+
if first_hit.reversed
|
402
|
+
rev_comp
|
403
|
+
first_hit.reversed=FALSE
|
404
|
+
end
|
405
|
+
end
|
406
|
+
|
407
|
+
def rev_comp
|
408
|
+
rev_seq=[]
|
409
|
+
@seq.each_char do |char|
|
410
|
+
char.upcase!
|
411
|
+
if char =='A'
|
412
|
+
rev_seq << 'T'
|
413
|
+
elsif char =='T'
|
414
|
+
rev_seq << 'A'
|
415
|
+
elsif char =='G'
|
416
|
+
rev_seq << 'C'
|
417
|
+
elsif char =='C'
|
418
|
+
rev_seq << 'G'
|
419
|
+
else
|
420
|
+
rev_seq << char
|
421
|
+
end
|
422
|
+
end
|
423
|
+
rev_seq.reverse!
|
424
|
+
@seq=rev_seq.join
|
425
|
+
|
426
|
+
end
|
427
|
+
|
428
|
+
def rev_coord
|
429
|
+
each_hit {|hit|
|
430
|
+
hit.rev_coord(@length)
|
431
|
+
hit.hsps_sort!
|
432
|
+
}
|
433
|
+
end
|
434
|
+
|
435
|
+
def start_codon_search #Busqueda codon inicio, busca si existe y una vez encontrado modifica coordenadas para acomodar el codon
|
436
|
+
exists=FALSE
|
437
|
+
s_beg=@hits.first.hsps.first.s_beg
|
438
|
+
s_end=@hits.first.hsps.first.s_end
|
439
|
+
q_beg=@hits.first.hsps.first.q_beg
|
440
|
+
q_end=@hits.first.hsps.first.q_end
|
441
|
+
if s_beg<=10 # Se busca codon si la proteina carece de los 10 primeros aa
|
442
|
+
continue=TRUE
|
443
|
+
index=0
|
444
|
+
temp_index=0
|
445
|
+
while continue==TRUE
|
446
|
+
if temp_index==0
|
447
|
+
find=@seq.index('ATG')
|
448
|
+
else
|
449
|
+
find=@seq.index('ATG',temp_index+1)
|
450
|
+
end
|
451
|
+
find+=1
|
452
|
+
if find==nil
|
453
|
+
continue=FALSE
|
454
|
+
else
|
455
|
+
if find==q_beg
|
456
|
+
continue=FALSE
|
457
|
+
index=find
|
458
|
+
elsif find>(q_beg-1+3)
|
459
|
+
continue=FALSE
|
460
|
+
else
|
461
|
+
if (find-q_end).modulo(3)==0 || find-q_end==0
|
462
|
+
index=find
|
463
|
+
end
|
464
|
+
if temp_index==0
|
465
|
+
temp_index=find+1
|
466
|
+
else
|
467
|
+
temp_index=find
|
468
|
+
end
|
469
|
+
end
|
470
|
+
end
|
471
|
+
end#While
|
472
|
+
if index>0
|
473
|
+
@hits.first.hsps.first.s_beg=1
|
474
|
+
@hits.first.hsps.first.q_beg=index
|
475
|
+
exists=TRUE
|
476
|
+
end
|
477
|
+
end
|
478
|
+
return exists
|
479
|
+
end
|
480
|
+
|
481
|
+
def stop_codon(codon,ends,*beg) #Busqueda posibles codones de parada
|
482
|
+
reference=ends-1
|
483
|
+
if !beg.empty?
|
484
|
+
reference=ends
|
485
|
+
ends=beg.first
|
486
|
+
end
|
487
|
+
position=nil
|
488
|
+
follow=TRUE
|
489
|
+
while follow
|
490
|
+
pos=@seq.index(codon,ends)#Implicitamente lleva el +1
|
491
|
+
if pos.nil?
|
492
|
+
follow=FALSE
|
493
|
+
else
|
494
|
+
dif=(pos-reference).abs
|
495
|
+
#puts "#{pos} #{dif} #{reference}"
|
496
|
+
if dif.modulo(3)==0
|
497
|
+
position=pos
|
498
|
+
follow=FALSE
|
499
|
+
else
|
500
|
+
ends=pos+1
|
501
|
+
end
|
502
|
+
end
|
503
|
+
end
|
504
|
+
return position
|
505
|
+
end
|
506
|
+
|
507
|
+
def coor_intrones
|
508
|
+
#Determinar bordes de los intrones
|
509
|
+
intrones=[]
|
510
|
+
last_hsp=nil
|
511
|
+
if first_hit.hsp_count>1
|
512
|
+
first_hit.each_hsp_with_index{|hsp,i|
|
513
|
+
if i>0
|
514
|
+
intrones << [last_hsp.q_end,hsp.q_beg]
|
515
|
+
end
|
516
|
+
last_hsp=hsp
|
517
|
+
}
|
518
|
+
end
|
519
|
+
#---------------------------------
|
520
|
+
return intrones
|
521
|
+
end
|
522
|
+
|
523
|
+
def stop_codon_search #Busqueda codon de parada, busca si existe
|
524
|
+
exists=FALSE
|
525
|
+
homology_start=first_hit.first_hsp.q_beg
|
526
|
+
homology_end=first_hit.last_hsp.q_end
|
527
|
+
n=1
|
528
|
+
codon=''
|
529
|
+
position=nil
|
530
|
+
intrones=coor_intrones
|
531
|
+
@seq.chars do |c|
|
532
|
+
if n<homology_start#Comenzar comparacion al principio del primer exon
|
533
|
+
n+=1
|
534
|
+
next
|
535
|
+
end
|
536
|
+
if n>homology_end#Terminar comparacion
|
537
|
+
break
|
538
|
+
end
|
539
|
+
#Saltar intrones
|
540
|
+
if !intrones.empty?
|
541
|
+
jump=FALSE
|
542
|
+
intrones.each do |int|
|
543
|
+
if n>int[0] && n<int[1] #NO se incluye el borde del exon
|
544
|
+
#print 'i'
|
545
|
+
jump=TRUE
|
546
|
+
break
|
547
|
+
end
|
548
|
+
end
|
549
|
+
if jump
|
550
|
+
n+=1
|
551
|
+
next
|
552
|
+
end
|
553
|
+
end
|
554
|
+
#-----------------
|
555
|
+
codon+=c
|
556
|
+
if codon.length==3 #Comparacion del codon
|
557
|
+
#puts position.to_s+"\t"+codon
|
558
|
+
if codon=='TAG'||codon=='TAA'||codon=='TGA'
|
559
|
+
@stops << position
|
560
|
+
exists=TRUE
|
561
|
+
end
|
562
|
+
codon=''
|
563
|
+
elsif codon.length==1 #Guardar posicion del primer nucleotido del codon
|
564
|
+
position=n
|
565
|
+
end
|
566
|
+
n+=1
|
567
|
+
end
|
568
|
+
return exists
|
569
|
+
end
|
570
|
+
|
571
|
+
def gff(id,parent,add) #Devuelve en estrutura gff los exones en genomico
|
572
|
+
text=[]
|
573
|
+
#Exones
|
574
|
+
first_hit.each_hsp{|hsp|
|
575
|
+
# if hsp.reversed==TRUE
|
576
|
+
# hsp.q_beg=@length-hsp.q_beg
|
577
|
+
# hsp.q_end=@length-hsp.q_end
|
578
|
+
# end
|
579
|
+
#puts "#{hsp.q_beg+add} #{hsp.q_end+add}"
|
580
|
+
text<<"#{id}\t.\texon\t#{hsp.q_beg+add}\t#{hsp.q_end+add}\t.\t+\t.\tID=#{parent}_exon;Parent=#{parent};Name=#{parent}_exon"
|
581
|
+
}
|
582
|
+
return text
|
583
|
+
end
|
584
|
+
|
585
|
+
def gff_prot(id,prot_name) #Devuelve en estrutura gff los exones en proteina
|
586
|
+
#parent="#{parent}_mRNA"
|
587
|
+
text=[]
|
588
|
+
#Exones
|
589
|
+
first_hit.each_hsp{|hsp|
|
590
|
+
#puts "#{hsp.q_beg+add} #{hsp.q_end+add}"
|
591
|
+
text<<"#{id}\t.\tprotein_match\t#{hsp.s_beg}\t#{hsp.s_end}\t.\t+\t.\tID=#{prot_name}_prot;Parent=#{prot_name};Name=#{id}_prot"
|
592
|
+
}
|
593
|
+
return text
|
594
|
+
end
|
595
|
+
|
596
|
+
def transfer_contig_hits(contig)
|
597
|
+
contig.each_hit{|hit|
|
598
|
+
self.transfer_hit(hit)
|
599
|
+
}
|
600
|
+
end
|
601
|
+
|
602
|
+
def transfer_hit(hit)
|
603
|
+
@hits << hit
|
604
|
+
end
|
605
|
+
|
606
|
+
def overlap
|
607
|
+
overlap=[]
|
608
|
+
last_hit=nil
|
609
|
+
each_hit_with_index{|hit,i|
|
610
|
+
overlap << hit.hsp_overlap
|
611
|
+
if i>0
|
612
|
+
diference=hit.overlap_with(last_hit)
|
613
|
+
if diference<0
|
614
|
+
overlap << diference
|
615
|
+
end
|
616
|
+
end
|
617
|
+
last_hit=hit
|
618
|
+
}
|
619
|
+
overlap.flatten!
|
620
|
+
return overlap
|
621
|
+
end
|
622
|
+
|
623
|
+
def hsp_at(position)
|
624
|
+
hsp_ret=nil
|
625
|
+
count_hsp=0
|
626
|
+
found=FALSE
|
627
|
+
each_hit {|hit|
|
628
|
+
hit.each_hsp {|hsp|
|
629
|
+
if position==count_hsp
|
630
|
+
hsp_ret=hsp
|
631
|
+
found=TRUE
|
632
|
+
break
|
633
|
+
end
|
634
|
+
count_hsp+=1
|
635
|
+
}
|
636
|
+
if found
|
637
|
+
break
|
638
|
+
end
|
639
|
+
}
|
640
|
+
return hsp_ret
|
641
|
+
end
|
642
|
+
|
643
|
+
end #class
|