ViewVC Help
View File | Revision Log | Show Annotations | View Changeset | Root Listing
root/gclib/gclib/gff.cpp
Revision: 197
Committed: Sun Feb 26 18:02:05 2012 UTC (7 years, 7 months ago) by gpertea
File size: 69639 byte(s)
Log Message:
trans.gff3 seems to work

Line File contents
1 #include "gff.h"
2
3 //GffNames* GffReader::names=NULL;
4 GffNames* GffObj::names=NULL;
5 //global set of feature names, attribute names etc.
6 // -- common for all GffObjs in current application!
7
8 const uint GFF_MAX_LOCUS = 7000000; //longest known gene in human is ~2.2M, UCSC claims a gene for mouse of ~ 3.1 M
9 const uint GFF_MAX_EXON = 30000; //longest known exon in human is ~11K
10 const uint GFF_MAX_INTRON= 6000000; //Ensembl shows a >5MB human intron
11 bool gff_show_warnings = false; //global setting, set by GffReader->showWarnings()
12 const int gff_fid_mRNA=0;
13 const int gff_fid_transcript=1;
14 const int gff_fid_exon=2;
15 const int gff_fid_CDS=3; //never really used in GffObj ftype_id or subftype_id
16 const uint gfo_flag_HAS_ERRORS = 0x00000001;
17 const uint gfo_flag_CHILDREN_PROMOTED= 0x00000002;
18 const uint gfo_flag_IS_GENE = 0x00000004;
19 const uint gfo_flag_IS_TRANSCRIPT = 0x00000008;
20 const uint gfo_flag_HAS_GFF_ID = 0x00000010; //found GFF3 feature line with its own ID
21 const uint gfo_flag_BY_EXON = 0x00000020; //created by subfeature (exon) directly
22 const uint gfo_flag_DISCARDED = 0x00000100;
23 const uint gfo_flag_LST_KEEP = 0x00000200;
24 const uint gfo_flag_LEVEL_MSK = 0x00FF0000;
25 const byte gfo_flagShift_LEVEL = 16;
26
27 void gffnames_ref(GffNames* &n) {
28 if (n==NULL) n=new GffNames();
29 n->numrefs++;
30 }
31
32 void gffnames_unref(GffNames* &n) {
33 if (n==NULL) GError("Error: attempt to remove reference to null GffNames object!\n");
34 n->numrefs--;
35 if (n->numrefs==0) { delete n; n=NULL; }
36 }
37
38 int gfo_cmpByLoc(const pointer p1, const pointer p2) {
39
40 GffObj& g1=*((GffObj*)p1);
41 GffObj& g2=*((GffObj*)p2);
42 if (g1.gseq_id==g2.gseq_id) {
43 if (g1.start!=g2.start)
44 return (int)(g1.start-g2.start);
45 else if (g1.getLevel()!=g2.getLevel())
46 return (int)(g1.getLevel()-g2.getLevel());
47 else
48 if (g1.end!=g2.end)
49 return (int)(g1.end-g2.end);
50 else return strcmp(g1.getID(), g2.getID());
51 }
52 else return (int)(g1.gseq_id-g2.gseq_id);
53 }
54
55 char* GffLine::extractAttr(const char* attr, bool caseStrict, bool enforce_GTF2) {
56 //parse a key attribute and remove it from the info string
57 //(only works for attributes that have values following them after ' ' or '=')
58 static const char GTF2_ERR[]="Error parsing attribute %s ('\"' required) at GTF line:\n%s\n";
59 int attrlen=strlen(attr);
60 char cend=attr[attrlen-1];
61 //char* pos = (caseStrict) ? strstr(info, attr) : strifind(info, attr);
62 //must make sure attr is not found in quoted text
63 char* pos=info;
64 char prevch=0;
65 bool in_str=false;
66 bool notfound=true;
67 int (*strcmpfn)(const char*, const char*, int) = caseStrict ? Gstrcmp : Gstricmp;
68 while (notfound && *pos) {
69 char ch=*pos;
70 if (ch=='"') {
71 in_str=!in_str;
72 pos++;
73 prevch=ch;
74 continue;
75 }
76 if (!in_str && (prevch==0 || prevch==' ' || prevch == ';')
77 && strcmpfn(attr, pos, attrlen)==0) {
78 //attr match found
79 //check for word boundary on right
80 char* epos=pos+attrlen;
81 if (cend=='=' || cend==' ' || *epos==0 || *epos==' ') {
82 notfound=false;
83 break;
84 }
85 //not a perfect match, move on
86 pos=epos;
87 prevch=*(pos-1);
88 continue;
89 }
90 //not a match or in_str
91 prevch=ch;
92 pos++;
93 }
94 if (notfound) return NULL;
95 char* vp=pos+attrlen;
96 while (*vp==' ') vp++;
97 if (*vp==';' || *vp==0)
98 GError("Error parsing value of GFF attribute \"%s\", line:\n%s\n", attr, dupline);
99 bool dq_enclosed=false; //value string enclosed by double quotes
100 if (*vp=='"') {
101 dq_enclosed=true;
102 vp++;
103 }
104 if (enforce_GTF2 && !dq_enclosed)
105 GError(GTF2_ERR,attr, dupline);
106 char* vend=vp;
107 if (dq_enclosed) {
108 while (*vend!='"' && *vend!=';' && *vend!=0) vend++;
109 }
110 else {
111 while (*vend!=';' && *vend!=0) vend++;
112 }
113 if (enforce_GTF2 && *vend!='"')
114 GError(GTF2_ERR, attr, dupline);
115 char *r=Gstrdup(vp, vend-1);
116 //-- now remove this attribute from the info string
117 while (*vend!=0 && (*vend=='"' || *vend==';' || *vend==' ')) vend++;
118 if (*vend==0) vend--;
119 for (char *src=vend, *dest=pos;;src++,dest++) {
120 *dest=*src;
121 if (*src==0) break;
122 }
123 return r;
124 }
125
126 static char fnamelc[128];
127
128 GffLine::GffLine(GffReader* reader, const char* l) {
129 llen=strlen(l);
130 GMALLOC(line,llen+1);
131 memcpy(line, l, llen+1);
132 GMALLOC(dupline, llen+1);
133 memcpy(dupline, l, llen+1);
134 skip=true;
135 gseqname=NULL;
136 track=NULL;
137 ftype=NULL;
138 info=NULL;
139 _parents=NULL;
140 _parents_len=0;
141 num_parents=0;
142 parents=NULL;
143 is_gff3=false;
144 is_cds=false;
145 is_transcript=false;
146 is_exon=false;
147 is_gene=false;
148 exontype=0;
149 gene_id=NULL;
150 gene_name=NULL;
151 qstart=0;
152 qend=0;
153 qlen=0;
154 ID=NULL;
155 char* t[9];
156 int i=0;
157 int tidx=1;
158 t[0]=line;
159
160 while (line[i]!=0) {
161 if (line[i]=='\t') {
162 line[i]=0;
163 t[tidx]=line+i+1;
164 tidx++;
165 if (tidx>8) break;
166 }
167 i++;
168 }
169
170 if (tidx<8) { // ignore non-GFF lines
171 // GMessage("Warning: error parsing GFF/GTF line:\n%s\n", l);
172 return;
173 }
174 gseqname=t[0];
175 track=t[1];
176 ftype=t[2];
177 info=t[8];
178 char* p=t[3];
179 if (!parseUInt(p,fstart)) {
180 //FIXME: chromosome_band entries in Flybase
181 GMessage("Warning: invalid start coordinate at line:\n%s\n",l);
182 return;
183 }
184 p=t[4];
185 if (!parseUInt(p,fend)) {
186 GMessage("Warning: invalid end coordinate at line:\n%s\n",l);
187 return;
188 }
189 if (fend<fstart) Gswap(fend,fstart); //make sure fstart>=fend, always
190 p=t[5];
191 if (p[0]=='.' && p[1]==0) {
192 score=0;
193 }
194 else {
195 if (!parseDouble(p,score))
196 GError("Error parsing feature score from GFF line:\n%s\n",l);
197 }
198 strand=*t[6];
199 if (strand!='+' && strand!='-' && strand!='.')
200 GError("Error parsing strand (%c) from GFF line:\n%s\n",strand,l);
201 phase=*t[7]; // must be '.', '0', '1' or '2'
202 ID=NULL;
203 // exon/CDS/mrna filter
204 strncpy(fnamelc, ftype, 127);
205 fnamelc[127]=0;
206 strlower(fnamelc); //convert to lower case
207 bool is_t_data=false;
208 if (strstr(fnamelc, "utr")!=NULL) {
209 exontype=exgffUTR;
210 is_exon=true;
211 is_t_data=true;
212 }
213 else if (endsWith(fnamelc, "exon")) {
214 exontype=exgffExon;
215 is_exon=true;
216 is_t_data=true;
217 }
218 else if (strstr(fnamelc, "stop") &&
219 (strstr(fnamelc, "codon") || strstr(fnamelc, "cds"))){
220 exontype=exgffStop;
221 is_cds=true; //though some place it outside the last CDS segment
222 is_t_data=true;
223 }
224 else if (strstr(fnamelc, "start") &&
225 ((strstr(fnamelc, "codon")!=NULL) || strstr(fnamelc, "cds")!=NULL)){
226 exontype=exgffStart;
227 is_cds=true;
228 is_t_data=true;
229 }
230 else if (strcmp(fnamelc, "cds")==0) {
231 exontype=exgffCDS;
232 is_cds=true;
233 is_t_data=true;
234 }
235 else if (endsWith(fnamelc, "gene") || startsWith(fnamelc, "gene")) {
236 is_gene=true;
237 is_t_data=true; //because its name will be attached to parented transcripts
238 }
239 else if (endsWith(fnamelc,"rna") || endsWith(fnamelc,"transcript")) {
240 is_transcript=true;
241 is_t_data=true;
242 }
243
244 if (reader->transcriptsOnly && !is_t_data) {
245 char* id=extractAttr("ID=");
246 if (id==NULL) id=extractAttr("transcript_id");
247 //GMessage("Discarding non-transcript line:\n%s\n",l);
248 if (id!=NULL) {
249 reader->discarded_ids.Add(id, new int(1));
250 GFREE(id);
251 }
252 return; //skip this line, unwanted feature name
253 }
254 ID=extractAttr("ID=",true);
255 char* Parent=extractAttr("Parent=",true);
256 is_gff3=(ID!=NULL || Parent!=NULL);
257 if (is_gff3) {
258 //parse as GFF3
259 if (ID!=NULL) {
260 //has ID attr so it's likely to be a parent feature
261 //look for explicit gene name
262 gene_name=extractAttr("gene_name=");
263 if (gene_name==NULL) {
264 gene_name=extractAttr("geneName=");
265 if (gene_name==NULL) {
266 gene_name=extractAttr("gene_sym=");
267 if (gene_name==NULL) {
268 gene_name=extractAttr("gene=");
269 }
270 }
271 }
272 gene_id=extractAttr("geneID=");
273 if (gene_id==NULL) {
274 gene_id=extractAttr("gene_id=");
275 }
276 if (is_gene) {
277 //special case: keep the Name and ID attributes of the gene feature
278 if (gene_name==NULL)
279 gene_name=extractAttr("Name=");
280 if (gene_id==NULL) //the ID is also gene_id in this case
281 gene_id=Gstrdup(ID);
282 //skip=false;
283 //return;
284 GFREE(Parent); //TMI, we really don't care about gene Parents?
285 } //gene feature
286 }// has GFF3 ID
287 if (Parent!=NULL) {
288 //keep Parent attr
289 //parse multiple parents
290 num_parents=1;
291 p=Parent;
292 int last_delim_pos=-1;
293 while (*p!=';' && *p!=0) {
294 if (*p==',' && *(p+1)!=0 && *(p+1)!=';') {
295 num_parents++;
296 last_delim_pos=(p-Parent);
297 }
298 p++;
299 }
300 _parents_len=p-Parent+1;
301 _parents=Parent;
302 GMALLOC(parents, num_parents*sizeof(char*));
303 parents[0]=_parents;
304 int i=1;
305 if (last_delim_pos>0) {
306 for (p=_parents+1;p<=_parents+last_delim_pos;p++) {
307 if (*p==',') {
308 char* ep=p-1;
309 while (*ep==' ' && ep>_parents) ep--;
310 *(ep+1)=0; //end the string there
311 parents[i]=p+1;
312 i++;
313 }
314 }
315 }
316 } //has Parent field
317 } //GFF3
318 else { // GTF-like expected
319 Parent=extractAttr("transcript_id",true);
320 if (Parent!=NULL) { //GTF2 format detected
321 if (is_transcript) {
322 // atypical GTF with a parent transcript line declared
323 ID=Parent;
324 Parent=NULL;
325 }
326 gene_id=extractAttr("gene_id"); // for GTF this is the only attribute accepted as geneID
327 if (gene_id==NULL)
328 gene_id=extractAttr("geneid");
329 gene_name=extractAttr("gene_name");
330 if (gene_name==NULL) {
331
332 gene_name=extractAttr("gene_sym");
333 if (gene_name==NULL) {
334 gene_name=extractAttr("gene");
335 if (gene_name==NULL)
336 gene_name=extractAttr("genesymbol");
337 }
338 }
339 //prepare for parseAttr by adding '=' character instead of spaces for all attributes
340 //after the attribute name
341 p=info;
342 bool noed=true; //not edited after the last delim
343 bool nsp=false; //non-space found after last delim
344 while (*p!=0) {
345 if (*p==' ') {
346 if (nsp && noed) {
347 *p='=';
348 noed=false;
349 p++;
350 continue;
351 }
352 }
353 else nsp=true; //non-space
354 if (*p==';') { noed=true; nsp=false; }
355 p++;
356 }
357 } //GTF2 detected (no parent line)
358 else {// Parent is NULL, check for jigsaw format or other pre-GTF2 format
359 //char* fexon=strstr(fnamelc, "exon");
360 //if (fexon!=NULL) {
361 if (exontype==exgffExon) {
362 if (startsWith(track,"jigsaw")) {
363 is_cds=true;
364 strcpy(track,"jigsaw");
365 p=strchr(info,';');
366 if (p==NULL) { Parent=Gstrdup(info); info=NULL; }
367 else { Parent=Gstrdup(info,p-1);
368 info=p+1;
369 }
370 }
371 } //exon feature?
372 if (Parent==NULL && exontype>=exgffCDS &&
373 (i=strcspn(info,"; \t\n\r"))<=(int)(strlen(info)+1)) {
374 //one word ID ? really desperate attempt to parse it here
375 Parent=Gstrdup(info,info+i-1);
376 info=NULL; //discard anything else on the line
377 }
378 }
379 if (Parent!=NULL) { //GTF transcript_id for exon/CDS feature
380 _parents=Parent;
381 GMALLOC(parents,sizeof(char*));
382 num_parents=1;
383 parents[0]=_parents;
384 }
385 } //GTF-like
386
387 //parse other potentially useful features
388 if (is_gff3) {
389 if ((p=strstr(info,"Target="))!=NULL) { //has Target attr
390 p+=7;
391 while (*p!=';' && *p!=0 && *p!=' ') p++;
392 if (*p!=' ') {
393 GError("Error parsing target coordinates from GFF line:\n%s\n",l);
394 }
395 if (!parseUInt(p,qstart))
396 GError("Error parsing target start coordinate from GFF line:\n%s\n",l);
397 if (*p!=' ') {
398 GError("Error parsing next target coordinate from GFF line:\n%s\n",l);
399 }
400 p++;
401 if (!parseUInt(p,qend))
402 GError("Error parsing target end coordinate from GFF line:\n%s\n",l);
403 }
404 if ((p=strifind(info,"Qreg="))!=NULL) { //has Qreg attr
405 p+=5;
406 if (!parseUInt(p,qstart))
407 GError("Error parsing target start coordinate from GFF line:\n%s\n",l);
408 if (*p!='-') {
409 GError("Error parsing next target coordinate from GFF line:\n%s\n",l);
410 }
411 p++;
412 if (!parseUInt(p,qend))
413 GError("Error parsing target end coordinate from GFF line:\n%s\n",l);
414 if (*p=='|' || *p==':') {
415 p++;
416 if (!parseUInt(p,qlen))
417 GError("Error parsing target length from GFF Qreg|: \n%s\n",l);
418 }
419 }//has Qreg attr
420 if (qlen==0 && (p=strifind(info,"Qlen="))!=NULL) {
421 p+=5;
422 if (!parseUInt(p,qlen))
423 GError("Error parsing target length from GFF Qlen:\n%s\n",l);
424 }
425 }//parsing some useful attributes in GFF3 records
426 if (ID==NULL && parents==NULL) {
427 if (reader->gff_warns)
428 GMessage("Warning: could not parse ID or Parent from GFF line:\n%s\n",dupline);
429 return; //skip
430 }
431 skip=false;
432 }
433
434
435 void GffObj::addCDS(uint cd_start, uint cd_end, char phase) {
436 if (cd_start>=this->start) {
437 this->CDstart=cd_start;
438 if (strand=='+') this->CDphase=phase;
439 }
440 else this->CDstart=this->start;
441 if (cd_end<=this->end) {
442 this->CDend=cd_end;
443 if (strand=='-') this->CDphase=phase;
444 }
445 else this->CDend=this->end;
446 isTranscript(true);
447 exon_ftype_id=gff_fid_exon;
448 if (monoFeature()) {
449 if (exons.Count()==0) addExon(this->start, this->end,0,'.',0,0,false,exgffExon);
450 else exons[0]->exontype=exgffExon;
451 }
452 }
453
454 int GffObj::addExon(GffReader* reader, GffLine* gl, bool keepAttr, bool noExonAttr) {
455 //this will make sure we have the right subftype_id!
456 //int subf_id=-1;
457 if (!isTranscript() && gl->is_cds) {
458 isTranscript(true);
459 exon_ftype_id=gff_fid_exon;
460 if (exons.Count()==1) exons[0]->exontype=exgffExon;
461 }
462 if (isTranscript()) {
463 if (exon_ftype_id<0) {//exon_ftype_id=gff_fid_exon;
464 if (gl->exontype>0) exon_ftype_id=gff_fid_exon;
465 else exon_ftype_id=names->feats.addName(gl->ftype);
466 }
467 //any recognized mRNA segment gets the generic "exon" type (also applies to CDS)
468 if (gl->exontype==0 && !gl->is_transcript) {
469 //extraneous mRNA feature, discard
470 if (reader->gff_warns)
471 GMessage("Warning: discarding unrecognized transcript subfeature %s of %s\n",
472 gl->ftype, gffID);
473 return -1;
474 }
475 }
476 else { //non-mRNA parent feature, check this subf type
477 int subf_id=names->feats.addName(gl->ftype);
478 if (exon_ftype_id<0 || exons.Count()==0) //never assigned a subfeature type before (e.g. first exon being added)
479 exon_ftype_id=subf_id;
480 else {
481 if (exon_ftype_id!=subf_id) {
482 //
483 if (exon_ftype_id==ftype_id && exons.Count()==1 && exons[0]->start==start && exons[0]->end==end) {
484 //the existing exon was just a dummy one created by default, discard it
485 exons.Clear();
486 covlen=0;
487 exon_ftype_id=subf_id; //allow the new subfeature to completely takeover
488 }
489 else { //multiple subfeatures, prefer those with
490 if (reader->gff_warns)
491 GMessage("GFF Warning: multiple subfeatures (%s and %s) found for %s, discarding ",
492 names->feats.getName(subf_id), names->feats.getName(exon_ftype_id),gffID);
493 if (gl->exontype!=0) { //new feature is an exon, discard previously parsed subfeatures
494 if (reader->gff_warns) GMessage("%s.\n", names->feats.getName(exon_ftype_id));
495 exon_ftype_id=subf_id;
496 exons.Clear();
497 covlen=0;
498 }
499 else { //discard new feature
500 if (reader->gff_warns) GMessage("%s.\n", names->feats.getName(subf_id));
501 return -1; //skip this 2nd subfeature type for this parent!
502 }
503 }
504 } //incoming subfeature is of different type
505 } //new subfeature type
506 } //non-mRNA parent
507 int eidx=addExon(gl->fstart, gl->fend, gl->score, gl->phase,
508 gl->qstart,gl->qend, gl->is_cds, gl->exontype);
509 if (eidx<0) return eidx; //this should never happen
510 if (keepAttr) {
511 if (noExonAttr) {
512 if (attrs==NULL) //place the parsed attributes directly at transcript level
513 parseAttrs(attrs, gl->info);
514 }
515 else { //need all exon-level attributes
516 parseAttrs(exons[eidx]->attrs, gl->info, true);
517 }
518 }
519 return eidx;
520 }
521
522
523 int GffObj::addExon(uint segstart, uint segend, double sc, char fr, int qs, int qe, bool iscds, char exontype) {
524 if (exons.Count()==0) {
525 if (iscds) isCDS=true; //for now, assume CDS only if first "exon" given is a CDS
526 if (exon_ftype_id<0) {
527 exon_ftype_id = isTranscript() ? gff_fid_exon : ftype_id;
528 }
529 }
530 //special treatment of start/stop codon features, they might be broken/split between exons
531 //and in that case some providers will still give the wrong end coordinate as start+2 (e.g. UCSC)
532 //so we should not trust the end coordinate for such features
533 if (exontype==exgffStart || exontype==exgffStop) {
534 if (strand=='-') segstart=segend;
535 else segend=segstart;
536 if (exontype==exgffStart) {
537 if (CDstart==0 || segstart<CDstart) CDstart=segstart;
538 }
539 else {
540 if (segstart>CDend) CDend=segstart;
541 }
542 }
543 else if (iscds) { //update CDS anchors:
544 if (CDstart==0 || segstart<CDstart) {
545 CDstart=segstart;
546 if (exontype==exgffCDS && strand=='+') CDphase=fr;
547 }
548 if (segend>CDend) {
549 if (exontype==exgffCDS && strand=='-') CDphase=fr;
550 CDend=segend;
551 }
552 }
553 else { // not a CDS/start/stop
554 isCDS=false;
555 }
556 if (qs || qe) {
557 if (qs>qe) Gswap(qs,qe);
558 if (qs==0) qs=1;
559 }
560 int ovlen=0;
561 if (exontype>0) { //check for overlaps between exon-type segments
562 int oi=exonOverlapIdx(segstart, segend, &ovlen);
563 if (oi>=0) { //overlap existing segment
564 if (ovlen==0) {
565 //adjacent segments will be merged
566 //e.g. CDS to (UTR|exon)
567 if ((exons[oi]->exontype>=exgffUTR && exontype==exgffCDS) ||
568 (exons[oi]->exontype==exgffCDS && exontype>=exgffUTR)) {
569 expandExon(oi, segstart, segend, exgffCDSUTR, sc, fr, qs, qe);
570 return oi;
571 }
572 //CDS adjacent to stop_codon: UCSC does (did?) this
573 if ((exons[oi]->exontype==exgffStop && exontype==exgffCDS) ||
574 (exons[oi]->exontype==exgffCDS && exontype==exgffStop)) {
575 expandExon(oi, segstart, segend, exgffCDS, sc, fr, qs, qe);
576 return oi;
577 }
578 }
579 //only allow this for CDS within exon, stop_codon within (CDS|UTR|exon),
580 // start_codon within (CDS|exon)
581 if (exons[oi]->exontype>exontype &&
582 exons[oi]->start<=segstart && exons[oi]->end>=segend &&
583 !(exons[oi]->exontype==exgffUTR && exontype==exgffCDS)) {
584 //larger segment given first, now the smaller included one is redundant
585 return oi; //only used to store attributes from current GffLine
586 }
587 if (exontype>exons[oi]->exontype &&
588 segstart<=exons[oi]->start && segend>=exons[oi]->end &&
589 !(exontype==exgffUTR && exons[oi]->exontype==exgffCDS)) {
590 //smaller segment given first, so we have to enlarge it
591 expandExon(oi, segstart, segend, exontype, sc, fr, qs, qe);
592 //this should also check for overlapping next exon (oi+1) ?
593 return oi;
594 }
595 //there is also the special case of "ribosomal slippage exception" (programmed frameshift)
596 //where two CDS segments may actually overlap for 1 or 2 bases, but there should be only one encompassing exon
597 //if (ovlen>2 || exons[oi]->exontype!=exgffCDS || exontype!=exgffCDS) {
598 // had to relax this because of some weird UCSC annotations with exons partially overlapping the CDS segments
599 /*
600 if (ovlen>2 && exons[oi]->exontype!=exgffUTR && exontype!=exgffUTR) {
601 if (gff_show_warnings)
602 GMessage("GFF Warning: discarding overlapping feature segment (%d-%d) (vs %d-%d (%s)) for GFF ID %s on %s\n",
603 segstart, segend, exons[oi]->start, exons[oi]->end, getSubfName(), gffID, getGSeqName());
604 hasErrors(true);
605 return -1; //segment NOT added
606 }
607 */
608
609 if ((ovlen>2 || ovlen==0) || exons[oi]->exontype!=exgffCDS || exontype!=exgffCDS) {
610 if (gff_show_warnings)
611 GMessage("GFF Warning: merging overlapping/adjacent feature segment (%d-%d) into (%d-%d) (%s) for GFF ID %s on %s\n",
612 segstart, segend, exons[oi]->start, exons[oi]->end, getSubfName(), gffID, getGSeqName());
613 expandExon(oi, segstart, segend, exontype, sc, fr, qs, qe);
614 return oi;
615 }
616 // else add the segment if the overlap is small and between two CDS segments
617 //TODO: we might want to add an attribute here with the slippage coordinate and size?
618 covlen-=ovlen;
619 }//overlap or adjacent to existing segment
620 } //check for overlap
621 // --- no overlap, or accepted micro-overlap (ribosomal slippage)
622 // create & add the new segment
623 /*
624 if (start>0 && exontype==exgffCDS && exons.Count()==0) {
625 //adding a CDS directly as the first subfeature of a declared parent
626 segstart=start;
627 segend=end;
628 }
629 */
630 GffExon* enew=new GffExon(segstart, segend, sc, fr, qs, qe, exontype);
631 int eidx=exons.Add(enew);
632 if (eidx<0) {
633 //this would actually be acceptable if the object is a "Gene" and "exons" are in fact isoforms
634 if (gff_show_warnings)
635 GMessage("GFF Warning: failed adding segment %d-%d for %s (discarded)!\n",
636 segstart, segend, gffID);
637 delete enew;
638 hasErrors(true);
639 return -1;
640 }
641 covlen+=(int)(exons[eidx]->end-exons[eidx]->start)+1;
642 //adjust parent feature coordinates to contain this exon
643 if (start==0 || start>exons.First()->start) {
644 start=exons.First()->start;
645 }
646 if (end<exons.Last()->end) end=exons.Last()->end;
647
648 if (uptr!=NULL) { //collect stats about the underlying genomic sequence
649 GSeqStat* gsd=(GSeqStat*)uptr;
650 if (start<gsd->mincoord) gsd->mincoord=start;
651 if (end>gsd->maxcoord) gsd->maxcoord=end;
652 if (this->len()>gsd->maxfeat_len) {
653 gsd->maxfeat_len=this->len();
654 gsd->maxfeat=this;
655 }
656 }
657 return eidx;
658 }
659
660 void GffObj::expandExon(int oi, uint segstart, uint segend, char exontype, double sc, char fr, int qs, int qe) {
661 //oi is the index of the *first* overlapping segment found that must be enlarged
662 covlen-=exons[oi]->len();
663 if (segstart<exons[oi]->start)
664 exons[oi]->start=segstart;
665 if (qs && qs<exons[oi]->qstart) exons[oi]->qstart=qs;
666 if (segend>exons[oi]->end)
667 exons[oi]->end=segend;
668 if (qe && qe>exons[oi]->qend) exons[oi]->qend=qe;
669 //warning: score cannot be properly adjusted! e.g. if it's a p-value it's just going to get worse
670 if (sc!=0) exons[oi]->score=sc;
671 covlen+=exons[oi]->len();
672 //if (exons[oi]->exontype< exontype) -- always true
673 exons[oi]->exontype = exontype;
674 if (exontype==exgffCDS) exons[oi]->phase=fr;
675 //we must check if any more exons are also overlapping this
676 int ni=oi+1; //next exon index after oi
677 while (ni<exons.Count() && segend>=exons[ni]->start) { // next segment overlaps new enlarged segment
678 //only allow this if next segment is fully included, and a subordinate
679 if (exons[ni]->exontype<exontype && exons[ni]->end<=segend) {
680 /* I guess we have to relax this due to stupid UCSC hg18 files having a start_codon sticking out
681 chr1 hg18_knownGene start_codon 69806911 69806913 0.000000 + .
682 chr1 hg18_knownGene CDS 69806911 69806912 0.000000 + 0
683 chr1 hg18_knownGene exon 69805456 69806912 0.000000 + .
684 */
685 if (exons[ni]->qstart<exons[oi]->qstart) exons[oi]->qstart=exons[ni]->qstart;
686 if (exons[ni]->qend>exons[oi]->qend) exons[oi]->qend=exons[ni]->qend;
687 exons.Delete(ni);
688 }
689 else {
690 if (gff_show_warnings) GMessage("GFF Warning: overlapping existing exon(%d-%d) while expanding to %d-%d for GFF ID %s\n",
691 exons[ni]->start, exons[ni]->end, segstart, segend, gffID);
692 //hasErrors(true);
693 break;
694 }
695 }
696 // -- make sure any other related boundaries are updated:
697 start=exons.First()->start;
698 end=exons.Last()->end;
699 if (uptr!=NULL) { //collect stats about the underlying genomic sequence
700 GSeqStat* gsd=(GSeqStat*)uptr;
701 if (start<gsd->mincoord) gsd->mincoord=start;
702 if (end>gsd->maxcoord) gsd->maxcoord=end;
703 if (this->len()>gsd->maxfeat_len) {
704 gsd->maxfeat_len=this->len();
705 gsd->maxfeat=this;
706 }
707 }
708 }
709
710 void GffObj::removeExon(int idx) {
711 /*
712 if (idx==0 && segs[0].start==gstart)
713 gstart=segs[1].start;
714 if (idx==segcount && segs[segcount].end==gend)
715 gend=segs[segcount-1].end;
716 */
717 if (idx<0 || idx>=exons.Count()) return;
718 int segstart=exons[idx]->start;
719 int segend=exons[idx]->end;
720 exons.Delete(idx);
721 covlen -= (int)(segend-segstart)+1;
722 start=exons.First()->start;
723 end=exons.Last()->end;
724 if (isCDS) { CDstart=start; CDend=end; }
725 }
726
727 void GffObj::removeExon(GffExon* p) {
728 for (int idx=0;idx<exons.Count();idx++) {
729 if (exons[idx]==p) {
730 int segstart=exons[idx]->start;
731 int segend=exons[idx]->end;
732 exons.Delete(idx);
733 covlen -= (int)(segend-segstart)+1;
734 start=exons.First()->start;
735 end=exons.Last()->end;
736 if (isCDS) { CDstart=start; CDend=end; }
737 return;
738 }
739 }
740 }
741
742
743
744 GffObj::GffObj(GffReader *gfrd, GffLine* gffline, bool keepAttr, bool noExonAttr):
745 GSeg(0,0), exons(true,true,false), children(1,false) {
746 xstart=0;
747 xend=0;
748 xstatus=0;
749 partial=false;
750 isCDS=false;
751 uptr=NULL;
752 ulink=NULL;
753 parent=NULL;
754 udata=0;
755 flags=0;
756 CDstart=0;
757 CDend=0;
758 CDphase=0;
759 geneID=NULL;
760 gene_name=NULL;
761 attrs=NULL;
762 gffID=NULL;
763 track_id=-1;
764 gseq_id=-1;
765 ftype_id=-1;
766 exon_ftype_id=-1;
767 strand='.';
768 if (gfrd==NULL)
769 GError("Cannot use this GffObj constructor with a NULL GffReader!\n");
770 gffnames_ref(names);
771 if (gfrd->names==NULL) gfrd->names=names;
772 //qlen=0;qstart=0;qend=0;
773 gscore=0;
774 uscore=0;
775 covlen=0;
776 qcov=0;
777 start=gffline->fstart;
778 end=gffline->fend;
779 gseq_id=names->gseqs.addName(gffline->gseqname);
780 track_id=names->tracks.addName(gffline->track);
781 strand=gffline->strand;
782 qlen=gffline->qlen;
783 qstart=gffline->qstart;
784 qend=gffline->qend;
785 //setup flags from gffline
786 isCDS=gffline->is_cds; //for now
787 isGene(gffline->is_gene);
788 isTranscript(gffline->is_transcript || gffline->exontype!=0);
789 //fromGff3(gffline->is_gff3);
790
791 if (gffline->parents!=NULL && !gffline->is_transcript) {
792 //GTF style -- create a GffObj directly by subfeature
793 //(also possible orphan GFF3 exon line, or an exon given before its parent (chado))
794 if (gffline->exontype!=0) { //recognized exon-like feature
795 ftype_id=gff_fid_transcript; //so this is some sort of transcript
796 exon_ftype_id=gff_fid_exon; //subfeatures MUST be exons
797 }
798 else {//unrecognized subfeatures
799 //make this GffObj of the same feature type
800 ftype_id=names->feats.addName(gffline->ftype);
801 }
802 if (gffline->ID==NULL) { //typical GTF2 without "transcript" line
803 gffID=Gstrdup(gffline->parents[0]);
804 this->createdByExon(true);
805 //this is likely the first exon/segment of the feature
806 addExon(gfrd, gffline, keepAttr, noExonAttr);
807 }
808 else { //a parented feature with an ID: orphan or premature GFF3 subfeature line
809 if (gffline->is_gff3 && gffline->exontype!=0) {
810 //premature exon given before its parent transcript
811 //create the transcript entry here
812 gffID=Gstrdup(gffline->parents[0]);
813 this->createdByExon(true);
814 //this is the first exon/segment of the transcript
815 addExon(gfrd, gffline, keepAttr, noExonAttr);
816 }
817 else { //unrecognized non-exon feature ? use the ID instead
818 this->hasGffID(true);
819 gffID=Gstrdup(gffline->ID);
820 if (keepAttr) this->parseAttrs(attrs, gffline->info);
821 }
822 }
823 } //non-transcript parented subfeature given directly
824 else {
825 //non-parented feature OR a recognizable transcript
826 //create a parent feature in its own right
827 gscore=gffline->score;
828 if (gffline->ID==NULL || gffline->ID[0]==0)
829 GError("Error: no ID found for GFF record start\n");
830 this->hasGffID(true);
831 gffID=Gstrdup(gffline->ID); //there must be an ID here
832 //if (gffline->is_transcript) ftype_id=gff_fid_mRNA;
833 //else
834 ftype_id=names->feats.addName(gffline->ftype);
835 if (gffline->is_transcript)
836 exon_ftype_id=gff_fid_exon;
837 if (keepAttr) this->parseAttrs(attrs, gffline->info);
838 }//no parent
839
840 if (gffline->gene_name!=NULL) {
841 gene_name=Gstrdup(gffline->gene_name);
842 }
843 if (gffline->gene_id) {
844 geneID=Gstrdup(gffline->gene_id);
845 }
846 else if (gffline->is_transcript && gffline->parents) {
847 geneID=Gstrdup(gffline->parents[0]);
848 }
849
850 GSeqStat* gsd=gfrd->gseqstats.AddIfNew(new GSeqStat(gseq_id,names->gseqs.lastNameUsed()),true);
851 uptr=gsd;
852 if (start<gsd->mincoord) gsd->mincoord=start;
853 if (end>gsd->maxcoord) gsd->maxcoord=end;
854 if (this->len()>gsd->maxfeat_len) {
855 gsd->maxfeat_len=this->len();
856 gsd->maxfeat=this;
857 }
858 }
859
860 GffLine* GffReader::nextGffLine() {
861 if (gffline!=NULL) return gffline; //caller should free gffline after processing
862 while (gffline==NULL) {
863 int llen=0;
864 buflen=GFF_LINELEN-1;
865 char* l=fgetline(linebuf, buflen, fh, &fpos, &llen);
866 if (l==NULL) {
867 return NULL; //end of file
868 }
869 int ns=0; //first nonspace position
870 while (l[ns]!=0 && isspace(l[ns])) ns++;
871 if (l[ns]=='#' || llen<10) continue;
872 gffline=new GffLine(this, l);
873 if (gffline->skip) {
874 delete gffline;
875 gffline=NULL;
876 continue;
877 }
878 if (gffline->ID==NULL && gffline->parents==NULL) { //it must have an ID
879 //this might not be needed, already checked in the GffLine constructor
880 if (gff_warns)
881 GMessage("Warning: malformed GFF line, no parent or record Id (kipping\n");
882 delete gffline;
883 gffline=NULL;
884 //continue;
885 }
886 }
887 return gffline;
888 }
889
890
891 char* GffReader::gfoBuildId(const char* id, const char* ctg) {
892 //caller must free the returned pointer
893 char* buf=NULL;
894 int idlen=strlen(id);
895 GMALLOC(buf, idlen+strlen(ctg)+2);
896 strcpy(buf, id);
897 buf[idlen]='~';
898 strcpy(buf+idlen+1, ctg);
899 return buf;
900 }
901 /*
902 void GffReader::gfoRemove(const char* id, const char* ctg) {
903 char* buf=gfoBuildId(id,ctg);
904 phash.Remove(buf);
905 GFREE(buf);
906 }
907 */
908 GfoHolder* GffReader::gfoAdd(GffObj* gfo, int idx) {
909 //Warning: if gflst gets altered, idx becomes obsolete
910 GVec<GfoHolder>* glst=phash.Find(gfo->gffID);
911 if (glst==NULL)
912 glst=new GVec<GfoHolder>(1);
913 GfoHolder gh(gfo,idx);
914 int i=glst->Add(gh);
915 phash.Add(gfo->gffID, glst);
916 return &(glst->Get(i));
917 }
918
919 GfoHolder* GffReader::gfoAdd(GVec<GfoHolder>& glst, GffObj* gfo, int idx) {
920 GfoHolder gh(gfo,idx);
921 int i=glst.Add(gh);
922 return &(glst[i]);
923 }
924
925 GfoHolder* GffReader::gfoFind(const char* id, const char* ctg,
926 GVec<GfoHolder>** glst, char strand, uint start, uint end) {
927 GVec<GfoHolder>* gl=phash.Find(id);
928 GfoHolder* gh=NULL;
929 if (gl) {
930 for (int i=0;i<gl->Count();i++) {
931 GfoHolder& gfo = gl->Get(i);
932 if (ctg!=NULL && strcmp(ctg, gfo.gffobj->getGSeqName())!=0)
933 continue;
934 if (strand && gfo.gffobj->strand!='.' && strand != gfo.gffobj->strand)
935 continue;
936 if (start>0) {
937 if (abs((int)start-(int)gfo.gffobj->start)>GFF_MAX_LOCUS)
938 continue;
939 if (end>0 && (gfo.gffobj->start>end || gfo.gffobj->end<start))
940 continue;
941 }
942 //must be the same transcript, according to given comparison criteria
943 gh=&gfo;
944 break;
945 }
946 }
947 if (glst) *glst=gl;
948 return gh;
949 }
950
951 GfoHolder* GffReader::replaceGffRec(GffLine* gffline, bool keepAttr, bool noExonAttr, int replaceidx) {
952 GffObj* newgfo=new GffObj(this, gffline, keepAttr, noExonAttr);
953 GfoHolder* r=NULL;
954 if (replaceidx>=0) {
955 gflst.Put(replaceidx,newgfo);
956 r=gfoAdd(newgfo, replaceidx);
957 }
958 else {
959 int gfoidx=gflst.Add(newgfo);
960 r=gfoAdd(newgfo, gfoidx);
961 }
962 /*
963 if (gff_warns) {
964 int* pcount=tids.Find(newgfo->gffID);
965 if (pcount!=NULL) {
966 if (gff_warns) GMessage("Warning: duplicate GFF ID: %s\n", newgfo->gffID);
967 (*pcount)++;
968 }
969 else {
970 tids.Add(newgfo->gffID,new int(1));
971 }
972 }
973 */
974 return r;
975 }
976
977 GfoHolder* GffReader::updateParent(GfoHolder* newgfh, GffObj* parent) {
978 //assert(parent);
979 //assert(newgfo);
980 parent->children.Add(newgfh->gffobj);
981 if (newgfh->gffobj->parent==NULL) newgfh->gffobj->parent=parent;
982 newgfh->gffobj->setLevel(parent->getLevel()+1);
983 if (parent->isGene()) {
984 if (parent->gene_name!=NULL && newgfh->gffobj->gene_name==NULL)
985 newgfh->gffobj->gene_name=Gstrdup(parent->gene_name);
986 if (parent->geneID!=NULL && newgfh->gffobj->geneID==NULL)
987 newgfh->gffobj->geneID=Gstrdup(parent->geneID);
988 }
989
990 return newgfh;
991 }
992
993 GfoHolder* GffReader::newGffRec(GffLine* gffline, bool keepAttr, bool noExonAttr,
994 GffObj* parent, GffExon* pexon, GVec<GfoHolder>* glst) {
995 GffObj* newgfo=new GffObj(this, gffline, keepAttr, noExonAttr);
996 GfoHolder* r=NULL;
997 int gfoidx=gflst.Add(newgfo);
998 r=(glst) ? gfoAdd(*glst, newgfo, gfoidx) : gfoAdd(newgfo, gfoidx);
999 if (parent!=NULL) {
1000 updateParent(r, parent);
1001 if (pexon!=NULL) parent->removeExon(pexon);
1002 }
1003 /*
1004 if (gff_warns) {
1005 int* pcount=tids.Find(newgfo->gffID);
1006 if (pcount!=NULL) {
1007 if (gff_warns) GMessage("Warning: duplicate GFF ID: %s\n", newgfo->gffID);
1008 (*pcount)++;
1009 }
1010 else {
1011 tids.Add(newgfo->gffID,new int(1));
1012 }
1013 }
1014 */
1015 return r;
1016 }
1017
1018 GfoHolder* GffReader::updateGffRec(GfoHolder* prevgfo, GffLine* gffline,
1019 bool keepAttr) {
1020 if (prevgfo==NULL) return NULL;
1021 //prevgfo->gffobj->createdByExon(false);
1022 prevgfo->gffobj->ftype_id=prevgfo->gffobj->names->feats.addName(gffline->ftype);
1023 prevgfo->gffobj->start=gffline->fstart;
1024 prevgfo->gffobj->end=gffline->fend;
1025 prevgfo->gffobj->isGene(gffline->is_gene);
1026 prevgfo->gffobj->isTranscript(gffline->is_transcript || gffline->exontype!=0);
1027 prevgfo->gffobj->hasGffID(gffline->ID!=NULL);
1028 if (keepAttr) {
1029 if (prevgfo->gffobj->attrs!=NULL) prevgfo->gffobj->attrs->Clear();
1030 prevgfo->gffobj->parseAttrs(prevgfo->gffobj->attrs, gffline->info);
1031 }
1032 return prevgfo;
1033 }
1034
1035
1036 bool GffReader::addExonFeature(GfoHolder* prevgfo, GffLine* gffline, GHash<CNonExon>& pex, bool noExonAttr) {
1037 bool r=true;
1038 if (gffline->strand!=prevgfo->gffobj->strand) {
1039 //TODO: add support for trans-splicing and even inter-chromosomal fusions
1040 if (prevgfo->gffobj->strand=='.') {
1041 prevgfo->gffobj->strand=gffline->strand;
1042 }
1043 else {
1044 GMessage("GFF Error at %s (%c): exon %d-%d (%c) found on different strand; discarded.\n",
1045 prevgfo->gffobj->gffID, prevgfo->gffobj->strand,
1046 gffline->fstart, gffline->fend, gffline->strand, prevgfo->gffobj->getGSeqName());
1047 //r=false;
1048 return true; //FIXME: split trans-spliced mRNAs by strand
1049 }
1050 }
1051 int gdist=(gffline->fstart>prevgfo->gffobj->end) ? gffline->fstart-prevgfo->gffobj->end :
1052 ((gffline->fend<prevgfo->gffobj->start)? prevgfo->gffobj->start-gffline->fend :
1053 0 );
1054 if (gdist>(int)GFF_MAX_LOCUS) { //too far apart, most likely this is a duplicate ID
1055 GMessage("Error: duplicate GFF ID '%s' (or exons too far apart)!\n",prevgfo->gffobj->gffID);
1056 //validation_errors = true;
1057 r=false;
1058 if (!gff_warns) exit(1);
1059 }
1060 int eidx=prevgfo->gffobj->addExon(this, gffline, !noExonAttr, noExonAttr);
1061 if (eidx>=0 && gffline->ID!=NULL && gffline->exontype==0)
1062 subfPoolAdd(pex, prevgfo);
1063 return r;
1064 }
1065
1066 CNonExon* GffReader::subfPoolCheck(GffLine* gffline, GHash<CNonExon>& pex, char*& subp_name) {
1067 CNonExon* subp=NULL;
1068 subp_name=NULL;
1069 for (int i=0;i<gffline->num_parents;i++) {
1070 if (transcriptsOnly && discarded_ids.Find(gffline->parents[i])!=NULL)
1071 continue;
1072 subp_name=gfoBuildId(gffline->parents[i], gffline->gseqname); //e.g. mRNA name
1073 subp=pex.Find(subp_name);
1074 if (subp!=NULL)
1075 return subp;
1076 GFREE(subp_name);
1077 }
1078 return NULL;
1079 }
1080
1081 void GffReader::subfPoolAdd(GHash<CNonExon>& pex, GfoHolder* newgfo) {
1082 //this might become a parent feature later
1083 if (newgfo->gffobj->exons.Count()>0) {
1084 char* xbuf=gfoBuildId(gffline->ID, gffline->gseqname);
1085 pex.Add(xbuf, new CNonExon(newgfo->idx, newgfo->gffobj,
1086 newgfo->gffobj->exons[0], gffline));
1087 GFREE(xbuf);
1088 }
1089 }
1090
1091 GfoHolder* GffReader::promoteFeature(CNonExon* subp, char*& subp_name, GHash<CNonExon>& pex,
1092 bool keepAttr, bool noExonAttr) {
1093 GffObj* prevp=subp->parent; //grandparent of gffline (e.g. gene)
1094 if (prevp!=gflst[subp->idx])
1095 GError("Error promoting subfeature %s, gflst index mismatch?!\n", subp->gffline->ID);
1096 subp->gffline->discardParent();
1097 GfoHolder* gfoh=newGffRec(subp->gffline, keepAttr, noExonAttr, prevp, subp->exon);
1098 pex.Remove(subp_name); //no longer a potential parent, moved it to phash already
1099 prevp->promotedChildren(true);
1100 return gfoh; //returns the holder of newly promoted feature
1101 }
1102
1103 //have to parse the whole file because exons can be scattered all over
1104 //trans-splicing and fusions are only accepted in proper GFF3 format, with a single parent feature ID entry
1105 void GffReader::readAll(bool keepAttr, bool mergeCloseExons, bool noExonAttr) {
1106 bool validation_errors = false;
1107 //loc_debug=false;
1108 GHash<CNonExon> pex; //keep track of any "exon"-like features that have an ID
1109 //and thus could become promoted to parent features
1110 while (nextGffLine()!=NULL) {
1111 GfoHolder* prevseen=NULL;
1112 GVec<GfoHolder>* prevgflst=NULL;
1113 if (gffline->ID && gffline->exontype==0) {
1114 //>> for a parent-like IDed feature (mRNA, gene, etc.)
1115 //look for same ID on the same chromosome/strand/locus
1116 prevseen=gfoFind(gffline->ID, gffline->gseqname, &prevgflst, gffline->strand, gffline->fstart);
1117 if (prevseen!=NULL) {
1118 //same ID/chromosome combo encountered before
1119 if (prevseen->gffobj->createdByExon() &&
1120 prevseen->gffobj->start>=gffline->fstart &&
1121 prevseen->gffobj->end<=gffline->fend) {
1122 //an exon of this ID was given before
1123 //this line has the main attributes for this ID
1124 updateGffRec(prevseen, gffline, keepAttr);
1125 }
1126 else {
1127 //- duplicate ID -- this must be a discontiguous feature
1128 // e.g. a trans-spliced transcript
1129 if (prevseen->gffobj->overlap(gffline->fstart, gffline->fend)) {
1130 //overlapping with same ID not allowed
1131 GMessage("Error: duplicate GFF ID '%s' encountered!\n",gffline->ID);
1132 //validation_errors = true;
1133 if (gff_warns) {
1134 delete gffline;
1135 gffline=NULL;
1136 continue;
1137 }
1138 else exit(1);
1139 }
1140 //create a new entry with the same ID
1141 prevseen=newGffRec(gffline, keepAttr, noExonAttr,
1142 prevseen->gffobj->parent, NULL, prevgflst);
1143 } //duplicate ID on the same chromosome
1144 } //prevseeen != NULL
1145 } //parent-like ID feature
1146 if (gffline->parents==NULL) {//start GFF3-like record with no parent (mRNA, gene)
1147 if (!prevseen) newGffRec(gffline, keepAttr, noExonAttr, NULL, NULL, prevgflst);
1148 }
1149 else { //--- it's a child feature (exon/CDS but could still be a mRNA with gene(s) as parent)
1150 //updates all the declared parents with this child
1151 bool found_parent=false;
1152 GfoHolder* newgfo=prevseen;
1153 GVec<GfoHolder>* newgflst=NULL;
1154 for (int i=0;i<gffline->num_parents;i++) {
1155 if (transcriptsOnly && discarded_ids.Find(gffline->parents[i])!=NULL)
1156 continue; //skipping discarded parent feature
1157 GfoHolder* parentgfo=NULL;
1158 if (gffline->is_transcript || gffline->exontype==0) {//possibly a transcript
1159 parentgfo=gfoFind(gffline->parents[i], gffline->gseqname,
1160 &newgflst, gffline->strand, gffline->fstart, gffline->fend);
1161 }
1162 else {
1163 //for exon-like entities we only need a parent to be in locus distance,
1164 //on the same strand
1165 parentgfo=gfoFind(gffline->parents[i], gffline->gseqname,
1166 &newgflst, gffline->strand, gffline->fstart);
1167 }
1168 if (parentgfo!=NULL) { //parent GffObj parsed earlier
1169 found_parent=true;
1170 if (parentgfo->gffobj->isGene() && gffline->is_transcript
1171 && gffline->exontype==0) {
1172 //not an exon, but a transcript parented by a gene
1173 if (newgfo) {
1174 updateParent(newgfo, parentgfo->gffobj);
1175 }
1176 else {
1177 newgfo=newGffRec(gffline, keepAttr, noExonAttr, parentgfo->gffobj);
1178 }
1179 }
1180 else { //potential exon subfeature
1181 if (!addExonFeature(parentgfo, gffline, pex, noExonAttr))
1182 validation_errors=true;
1183 }
1184 } //overlapping parent feature found
1185 } //for each parsed parent Id
1186 if (!found_parent) { //new GTF-like record starting here with a subfeature directly
1187 //or it could be some chado GFF3 barf with exons coming BEFORE their parent :(
1188 //check if this feature isn't parented by a previously stored "exon" subfeature
1189 char* subp_name=NULL;
1190 CNonExon* subp=subfPoolCheck(gffline, pex, subp_name);
1191 if (subp!=NULL) { //found a subfeature that is the parent of this gffline
1192 //promote that subfeature to a full GffObj
1193 GfoHolder* gfoh=promoteFeature(subp, subp_name, pex, keepAttr, noExonAttr);
1194 //add current gffline as an exon of the newly promoted subfeature
1195 if (!addExonFeature(gfoh, gffline, pex, noExonAttr))
1196 validation_errors=true;
1197 }
1198 else { //no parent seen before,
1199 //loc_debug=true;
1200 GfoHolder* ngfo=prevseen;
1201 if (ngfo==NULL) {
1202 //if it's an exon type, create directly the parent with this exon
1203 //but if it's recognized as a transcript, the object itself is created
1204 ngfo=newGffRec(gffline, keepAttr, noExonAttr, NULL, NULL, newgflst);
1205 }
1206 if (!ngfo->gffobj->isTranscript() &&
1207 gffline->ID!=NULL && gffline->exontype==0)
1208 subfPoolAdd(pex, ngfo);
1209 //even those with errors will be added here!
1210 }
1211 GFREE(subp_name);
1212 } //no previous parent found
1213 } //parented feature
1214 //--
1215 delete gffline;
1216 gffline=NULL;
1217 }//while gff lines
1218 gflst.finalize(this, mergeCloseExons, keepAttr, noExonAttr); //force sorting by locus if so constructed
1219 // all gff records are now loaded in GList gflst
1220 // so we can free the hash
1221 phash.Clear();
1222 //tids.Clear();
1223 if (validation_errors) {
1224 exit(1);
1225 }
1226 }
1227
1228 GffObj* GffObj::finalize(GffReader* gfr, bool mergeCloseExons, bool keepAttrs, bool noExonAttr) {
1229 //merge
1230 //always merge adjacent or overlapping segments
1231 //but if mergeCloseExons then merge even when distance is up to 5 bases
1232 udata=0;
1233 uptr=NULL;
1234 if (gfr->transcriptsOnly && !(isTranscript() || (isGene() && children.Count()==0))) {
1235 isDiscarded(true);
1236 }
1237 if (ftype_id==gff_fid_transcript && CDstart>0) {
1238 ftype_id=gff_fid_mRNA;
1239 //exon_ftype_id=gff_fid_exon;
1240 }
1241 if (exons.Count()>0 && (isTranscript() || exon_ftype_id==gff_fid_exon)) {
1242 if (mergeCloseExons) {
1243 int mindist=mergeCloseExons ? 5:1;
1244 for (int i=0;i<exons.Count()-1;i++) {
1245 int ni=i+1;
1246 uint mend=exons[i]->end;
1247 while (ni<exons.Count()) {
1248 int dist=(int)(exons[ni]->start-mend);
1249 if (dist>mindist) break; //no merging with next segment
1250 if (gfr!=NULL && gfr->gff_warns && dist!=0 && (exons[ni]->exontype!=exgffUTR && exons[i]->exontype!=exgffUTR)) {
1251 GMessage("GFF warning: merging adjacent/overlapping segments of %s on %s (%d-%d, %d-%d)\n",
1252 gffID, getGSeqName(), exons[i]->start, exons[i]->end,exons[ni]->start, exons[ni]->end);
1253 }
1254 mend=exons[ni]->end;
1255 covlen-=exons[i]->len();
1256 exons[i]->end=mend;
1257 covlen+=exons[i]->len();
1258 covlen-=exons[ni]->len();
1259 if (exons[ni]->attrs!=NULL && (exons[i]->attrs==NULL ||
1260 exons[i]->attrs->Count()<exons[ni]->attrs->Count())) {
1261 //use the other exon attributes, if more
1262 delete(exons[i]->attrs);
1263 exons[i]->attrs=exons[ni]->attrs;
1264 exons[ni]->attrs=NULL;
1265 }
1266 exons.Delete(ni);
1267 } //check for merge with next exon
1268 } //for each exon
1269 } //merge close exons
1270 //shrink transcript to the exons' span
1271 this->start=exons.First()->start;
1272 this->end=exons.Last()->end;
1273 }
1274 //attribute reduction for GTF records
1275 if (keepAttrs && !noExonAttr && !hasGffID()
1276 && exons.Count()>0 && exons[0]->attrs!=NULL) {
1277 bool attrs_discarded=false;
1278 for (int a=0;a<exons[0]->attrs->Count();a++) {
1279 int attr_name_id=exons[0]->attrs->Get(a)->attr_id;
1280 char* attr_name=names->attrs.getName(attr_name_id);
1281 char* attr_val =exons[0]->attrs->Get(a)->attr_val;
1282 bool sameExonAttr=true;
1283 for (int i=1;i<exons.Count();i++) {
1284 char* ov=exons[i]->getAttr(attr_name_id);
1285 if (ov==NULL || (strcmp(ov,attr_val)!=0)) {
1286 sameExonAttr=false;
1287 break;
1288 }
1289 }
1290 if (sameExonAttr) {
1291 //delete this attribute from exons level
1292 attrs_discarded=true;
1293 this->addAttr(attr_name, attr_val);
1294 for (int i=1;i<exons.Count();i++) {
1295 removeExonAttr(*(exons[i]), attr_name_id);
1296 }
1297 exons[0]->attrs->freeItem(a);
1298 }
1299 }
1300 if (attrs_discarded) exons[0]->attrs->Pack();
1301 }
1302 return this;
1303 }
1304
1305 void GffObj::parseAttrs(GffAttrs*& atrlist, char* info, bool isExon) {
1306 if (names==NULL)
1307 GError(ERR_NULL_GFNAMES, "parseAttrs()");
1308 if (atrlist==NULL)
1309 atrlist=new GffAttrs();
1310 char* endinfo=info+strlen(info);
1311 char* start=info;
1312 char* pch=start;
1313 while (start<endinfo) {
1314 //skip spaces
1315 while (*start==' ' && start<endinfo) start++;
1316 pch=strchr(start, ';');
1317 if (pch==NULL) pch=endinfo;
1318 else {
1319 *pch='\0';
1320 pch++;
1321 }
1322 char* ech=strchr(start,'=');
1323 if (ech!=NULL) { // attr=value format found
1324 *ech='\0';
1325 //if (noExonAttr && (strcmp(start, "exon_number")==0 || strcmp(start, "exon")==0)) { start=pch; continue; }
1326 if (strcmp(start, "exon_number")==0 || strcmp(start, "exon")==0 ||
1327 strcmp(start, "exon_id")==0)
1328 { start=pch; continue; }
1329 ech++;
1330 while (*ech==' ' && ech<endinfo) ech++;//skip extra spaces after the '='
1331 //atrlist->Add(new GffAttr(names->attrs.addName(start),ech));
1332 //make sure we don't add the same attribute more than once
1333 if (isExon && (strcmp(start, "protein_id")==0)) {
1334 //Ensembl special case
1335 this->addAttr(start, ech);
1336 start=pch;
1337 continue;
1338 }
1339 atrlist->add_or_update(names, start, ech);
1340 }
1341 /*
1342 else { //not an attr=value format
1343 atrlist->Add(new GffAttr(names->attrs.addName(start),"1"));
1344 }
1345 */
1346 start=pch;
1347 }
1348 if (atrlist->Count()==0) { delete atrlist; atrlist=NULL; }
1349 }
1350
1351 void GffObj::addAttr(const char* attrname, const char* attrvalue) {
1352 if (this->attrs==NULL)
1353 this->attrs=new GffAttrs();
1354 //this->attrs->Add(new GffAttr(names->attrs.addName(attrname),attrvalue));
1355 this->attrs->add_or_update(names, attrname, attrvalue);
1356 }
1357
1358
1359 void GffObj::setFeatureName(const char* feature) {
1360 //change the feature name/type for a transcript
1361 int fid=names->feats.addName(feature);
1362 if (monoFeature() && exons.Count()>0)
1363 this->exon_ftype_id=fid;
1364 this->ftype_id=fid;
1365 }
1366
1367 void GffObj::setRefName(const char* newname) {
1368 //change the feature name/type for a transcript
1369 int rid=names->gseqs.addName(newname);
1370 this->gseq_id=rid;
1371 }
1372
1373
1374
1375 int GffObj::removeAttr(const char* attrname, const char* attrval) {
1376 if (this->attrs==NULL || attrname==NULL || attrname[0]==0) return 0;
1377 int aid=this->names->attrs.getId(attrname);
1378 if (aid<0) return 0;
1379 int delcount=0; //could be more than one ?
1380 for (int i=0;i<this->attrs->Count();i++) {
1381 if (aid==this->attrs->Get(i)->attr_id) {
1382 if (attrval==NULL ||
1383 strcmp(attrval, this->attrs->Get(i)->attr_val)==0) {
1384 delcount++;
1385 this->attrs->freeItem(i);
1386 }
1387 }
1388 }
1389 if (delcount>0) this->attrs->Pack();
1390 return delcount;
1391 }
1392
1393 int GffObj::removeAttr(int aid, const char* attrval) {
1394 if (this->attrs==NULL || aid<0) return 0;
1395 int delcount=0; //could be more than one ?
1396 for (int i=0;i<this->attrs->Count();i++) {
1397 if (aid==this->attrs->Get(i)->attr_id) {
1398 if (attrval==NULL ||
1399 strcmp(attrval, this->attrs->Get(i)->attr_val)==0) {
1400 delcount++;
1401 this->attrs->freeItem(i);
1402 }
1403 }
1404 }
1405 if (delcount>0) this->attrs->Pack();
1406 return delcount;
1407 }
1408
1409
1410 int GffObj::removeExonAttr(GffExon& exon, const char* attrname, const char* attrval) {
1411 if (exon.attrs==NULL || attrname==NULL || attrname[0]==0) return 0;
1412 int aid=this->names->attrs.getId(attrname);
1413 if (aid<0) return 0;
1414 int delcount=0; //could be more than one
1415 for (int i=0;i<exon.attrs->Count();i++) {
1416 if (aid==exon.attrs->Get(i)->attr_id) {
1417 if (attrval==NULL ||
1418 strcmp(attrval, exon.attrs->Get(i)->attr_val)==0) {
1419 delcount++;
1420 exon.attrs->freeItem(i);
1421 }
1422 }
1423 }
1424 if (delcount>0) exon.attrs->Pack();
1425 return delcount;
1426 }
1427
1428 int GffObj::removeExonAttr(GffExon& exon, int aid, const char* attrval) {
1429 if (exon.attrs==NULL || aid<0) return 0;
1430 int delcount=0; //could be more than one
1431 for (int i=0;i<exon.attrs->Count();i++) {
1432 if (aid==exon.attrs->Get(i)->attr_id) {
1433 if (attrval==NULL ||
1434 strcmp(attrval, exon.attrs->Get(i)->attr_val)==0) {
1435 delcount++;
1436 exon.attrs->freeItem(i);
1437 }
1438 }
1439 }
1440 if (delcount>0) exon.attrs->Pack();
1441 return delcount;
1442 }
1443
1444
1445 void GffObj::getCDS_ends(uint& cds_start, uint& cds_end) {
1446 cds_start=0;
1447 cds_end=0;
1448 if (CDstart==0 || CDend==0) return; //no CDS info
1449 int cdsadj=0;
1450 if (CDphase=='1' || CDphase=='2') {
1451 cdsadj=CDphase-'0';
1452 }
1453 cds_start=CDstart;
1454 cds_end=CDend;
1455 if (strand=='-') cds_end-=cdsadj;
1456 else cds_start+=cdsadj;
1457 }
1458
1459 void GffObj::mRNA_CDS_coords(uint& cds_mstart, uint& cds_mend) {
1460 //sets cds_start and cds_end to the CDS start,end coordinates on the spliced mRNA transcript
1461 cds_mstart=0;
1462 cds_mend=0;
1463 if (CDstart==0 || CDend==0) return; //no CDS info
1464 //restore normal coordinates, just in case
1465 unxcoord();
1466 int cdsadj=0;
1467 if (CDphase=='1' || CDphase=='2') {
1468 cdsadj=CDphase-'0';
1469 }
1470 /*
1471 uint seqstart=CDstart;
1472 uint seqend=CDend;
1473 */
1474 uint seqstart=exons.First()->start;
1475 uint seqend=exons.Last()->end;
1476 int s=0; //resulting nucleotide counter
1477 if (strand=='-') {
1478 for (int x=exons.Count()-1;x>=0;x--) {
1479 uint sgstart=exons[x]->start;
1480 uint sgend=exons[x]->end;
1481 if (seqend<sgstart || seqstart>sgend) continue;
1482 if (seqstart>=sgstart && seqstart<=sgend)
1483 sgstart=seqstart; //seqstart within this segment
1484 if (seqend>=sgstart && seqend<=sgend)
1485 sgend=seqend; //seqend within this segment
1486 s+=(int)(sgend-sgstart)+1;
1487 if (CDstart>=sgstart && CDstart<=sgend) {
1488 //CDstart in this segment
1489 //and we are getting the whole transcript
1490 cds_mend=s-(int)(CDstart-sgstart);
1491 }
1492 if (CDend>=sgstart && CDend<=sgend) {
1493 //CDstart in this segment
1494 //and we are getting the whole transcript
1495 cds_mstart=s-(int)(CDend-cdsadj-sgstart);
1496 }
1497 } //for each exon
1498 } // - strand
1499 else { // + strand
1500 for (int x=0;x<exons.Count();x++) {
1501 uint sgstart=exons[x]->start;
1502 uint sgend=exons[x]->end;
1503 if (seqend<sgstart || seqstart>sgend) continue;
1504 if (seqstart>=sgstart && seqstart<=sgend)
1505 sgstart=seqstart; //seqstart within this segment
1506 if (seqend>=sgstart && seqend<=sgend)
1507 sgend=seqend; //seqend within this segment
1508 s+=(int)(sgend-sgstart)+1;
1509 /* for (uint i=sgstart;i<=sgend;i++) {
1510 spliced[s]=gsubseq[i-gstart];
1511 s++;
1512 }//for each nt
1513 */
1514 if (CDstart>=sgstart && CDstart<=sgend) {
1515 //CDstart in this segment
1516 cds_mstart=s-(int)(sgend-CDstart-cdsadj);
1517 }
1518 if (CDend>=sgstart && CDend<=sgend) {
1519 //CDend in this segment
1520 cds_mend=s-(int)(sgend-CDend);
1521 }
1522 } //for each exon
1523 } // + strand
1524 //spliced[s]=0;
1525 //if (rlen!=NULL) *rlen=s;
1526 //return spliced;
1527 }
1528
1529 char* GffObj::getUnspliced(GFaSeqGet* faseq, int* rlen, GList<GSeg>* seglst)
1530 {
1531 if (faseq==NULL) { GMessage("Warning: getUnspliced(NULL,.. ) called!\n");
1532 return NULL;
1533 }
1534 //restore normal coordinates:
1535 unxcoord();
1536 if (exons.Count()==0) return NULL;
1537 int fspan=end-start+1;
1538 const char* gsubseq=faseq->subseq(start, fspan);
1539 if (gsubseq==NULL) {
1540 GError("Error getting subseq for %s (%d..%d)!\n", gffID, start, end);
1541 }
1542 char* unspliced=NULL;
1543
1544 int seqstart=exons.First()->start;
1545 int seqend=exons.Last()->end;
1546
1547 int unsplicedlen = 0;
1548
1549 unsplicedlen += seqend - seqstart + 1;
1550
1551 GMALLOC(unspliced, unsplicedlen+1); //allocate more here
1552 //uint seqstart, seqend;
1553
1554 int s = 0; //resulting nucleotide counter
1555 if (strand=='-')
1556 {
1557 if (seglst!=NULL)
1558 seglst->Add(new GSeg(s+1,s+1+seqend-seqstart));
1559 for (int i=seqend;i>=seqstart;i--)
1560 {
1561 unspliced[s] = ntComplement(gsubseq[i-start]);
1562 s++;
1563 }//for each nt
1564 } // - strand
1565 else
1566 { // + strand
1567 if (seglst!=NULL)
1568 seglst->Add(new GSeg(s+1,s+1+seqend-seqstart));
1569 for (int i=seqstart;i<=seqend;i++)
1570 {
1571 unspliced[s]=gsubseq[i-start];
1572 s++;
1573 }//for each nt
1574 } // + strand
1575 //assert(s <= unsplicedlen);
1576 unspliced[s]=0;
1577 if (rlen!=NULL) *rlen=s;
1578 return unspliced;
1579 }
1580
1581 char* GffObj::getSpliced(GFaSeqGet* faseq, bool CDSonly, int* rlen, uint* cds_start, uint* cds_end,
1582 GList<GSeg>* seglst) {
1583 if (CDSonly && CDstart==0) return NULL;
1584 if (faseq==NULL) { GMessage("Warning: getSpliced(NULL,.. ) called!\n");
1585 return NULL;
1586 }
1587 //restore normal coordinates:
1588 unxcoord();
1589 if (exons.Count()==0) return NULL;
1590 int fspan=end-start+1;
1591 const char* gsubseq=faseq->subseq(start, fspan);
1592 if (gsubseq==NULL) {
1593 GError("Error getting subseq for %s (%d..%d)!\n", gffID, start, end);
1594 }
1595 if (fspan<(int)(end-start+1)) { //special case: stop coordinate was extended past the gseq length, must adjust
1596 int endadj=end-start+1-fspan;
1597 uint prevend=end;
1598 end-=endadj;
1599 if (CDend>end) CDend=end;
1600 if (exons.Last()->end>end) {
1601 exons.Last()->end=end; //this could get us into trouble if exon start is also > end
1602 if (exons.Last()->start>exons.Last()->end) {
1603 GError("GffObj::getSpliced() error: improper genomic coordinate %d on %s for %s\n",
1604 prevend,getGSeqName(), getID());
1605 }
1606 covlen-=endadj;
1607 }
1608 }
1609 char* spliced=NULL;
1610 GMALLOC(spliced, covlen+1); //allocate more here
1611 uint seqstart, seqend;
1612 int cdsadj=0;
1613 if (CDphase=='1' || CDphase=='2') {
1614 cdsadj=CDphase-'0';
1615 }
1616 if (CDSonly) {
1617 seqstart=CDstart;
1618 seqend=CDend;
1619 if (strand=='-') seqend-=cdsadj;
1620 else seqstart+=cdsadj;
1621 }
1622 else {
1623 seqstart=exons.First()->start;
1624 seqend=exons.Last()->end;
1625 }
1626 int s=0; //resulting nucleotide counter
1627 if (strand=='-') {
1628 for (int x=exons.Count()-1;x>=0;x--) {
1629 uint sgstart=exons[x]->start;
1630 uint sgend=exons[x]->end;
1631 if (seqend<sgstart || seqstart>sgend) continue;
1632 if (seqstart>=sgstart && seqstart<=sgend)
1633 sgstart=seqstart; //seqstart within this segment
1634 if (seqend>=sgstart && seqend<=sgend)
1635 sgend=seqend; //seqend within this segment
1636 if (seglst!=NULL)
1637 seglst->Add(new GSeg(s+1,s+1+sgend-sgstart));
1638 for (uint i=sgend;i>=sgstart;i--) {
1639 spliced[s] = ntComplement(gsubseq[i-start]);
1640 s++;
1641 }//for each nt
1642
1643 if (!CDSonly && cds_start!=NULL && CDstart>0) {
1644 if (CDstart>=sgstart && CDstart<=sgend) {
1645 //CDstart in this segment
1646 //and we are getting the whole transcript
1647 *cds_end=s-(CDstart-sgstart);
1648 }
1649 if (CDend>=sgstart && CDend<=sgend) {
1650 //CDstart in this segment
1651 //and we are getting the whole transcript
1652 *cds_start=s-(CDend-cdsadj-sgstart);
1653 }
1654 }//update local CDS coordinates
1655 } //for each exon
1656 } // - strand
1657 else { // + strand
1658 for (int x=0;x<exons.Count();x++) {
1659 uint sgstart=exons[x]->start;
1660 uint sgend=exons[x]->end;
1661 if (seqend<sgstart || seqstart>sgend) continue;
1662 if (seqstart>=sgstart && seqstart<=sgend)
1663 sgstart=seqstart; //seqstart within this segment
1664 if (seqend>=sgstart && seqend<=sgend)
1665 sgend=seqend; //seqend within this segment
1666 if (seglst!=NULL)
1667 seglst->Add(new GSeg(s+1,s+1+sgend-sgstart));
1668 for (uint i=sgstart;i<=sgend;i++) {
1669 spliced[s]=gsubseq[i-start];
1670 s++;
1671 }//for each nt
1672 if (!CDSonly && cds_start!=NULL && CDstart>0) {
1673 if (CDstart>=sgstart && CDstart<=sgend) {
1674 //CDstart in this segment
1675 //and we are getting the whole transcript
1676 *cds_start=s-(sgend-CDstart-cdsadj);
1677 }
1678 if (CDend>=sgstart && CDend<=sgend) {
1679 //CDstart in this segment
1680 //and we are getting the whole transcript
1681 *cds_end=s-(sgend-CDend);
1682 }
1683 }//update local CDS coordinates
1684 } //for each exon
1685 } // + strand
1686 spliced[s]=0;
1687 if (rlen!=NULL) *rlen=s;
1688 return spliced;
1689 }
1690
1691 char* GffObj::getSplicedTr(GFaSeqGet* faseq, bool CDSonly, int* rlen) {
1692 if (CDSonly && CDstart==0) return NULL;
1693 //restore normal coordinates:
1694 unxcoord();
1695 if (exons.Count()==0) return NULL;
1696 int fspan=end-start+1;
1697 const char* gsubseq=faseq->subseq(start, fspan);
1698 if (gsubseq==NULL) {
1699 GError("Error getting subseq for %s (%d..%d)!\n", gffID, start, end);
1700 }
1701
1702 char* translation=NULL;
1703 GMALLOC(translation, (int)(covlen/3)+1);
1704 uint seqstart, seqend;
1705 int cdsadj=0;
1706 if (CDphase=='1' || CDphase=='2') {
1707 cdsadj=CDphase-'0';
1708 }
1709 if (CDSonly) {
1710 seqstart=CDstart;
1711 seqend=CDend;
1712 if (strand=='-') seqend-=cdsadj;
1713 else seqstart+=cdsadj;
1714 }
1715 else {
1716 seqstart=exons.First()->start;
1717 seqend=exons.Last()->end;
1718 }
1719 Codon codon;
1720 int nt=0; //codon nucleotide counter (0..2)
1721 int aa=0; //aminoacid count
1722 if (strand=='-') {
1723 for (int x=exons.Count()-1;x>=0;x--) {
1724 uint sgstart=exons[x]->start;
1725 uint sgend=exons[x]->end;
1726 if (seqend<sgstart || seqstart>sgend) continue;
1727 if (seqstart>=sgstart && seqstart<=sgend)
1728 sgstart=seqstart; //seqstart within this segment
1729 if (seqend>=sgstart && seqend<=sgend) {
1730 sgend=seqend; //seqend within this segment
1731 }
1732 for (uint i=sgend;i>=sgstart;i--) {
1733 codon.nuc[nt]=ntComplement(gsubseq[i-start]);
1734 nt++;
1735 if (nt==3) {
1736 nt=0;
1737 translation[aa]=codon.translate();
1738 aa++;
1739 }
1740 }//for each nt
1741 } //for each exon
1742 } // - strand
1743 else { // + strand
1744 for (int x=0;x<exons.Count();x++) {
1745 uint sgstart=exons[x]->start;
1746 uint sgend=exons[x]->end;
1747 if (seqend<sgstart || seqstart>sgend) continue;
1748 if (seqstart>=sgstart && seqstart<=sgend)
1749 sgstart=seqstart; //seqstart within this segment
1750 if (seqend>=sgstart && seqend<=sgend)
1751 sgend=seqend; //seqend within this segment
1752 for (uint i=sgstart;i<=sgend;i++) {
1753 codon.nuc[nt]=gsubseq[i-start];
1754 nt++;
1755 if (nt==3) {
1756 nt=0;
1757 translation[aa]=codon.translate();
1758 aa++;
1759 }
1760 }//for each nt
1761 } //for each exon
1762 } // + strand
1763 translation[aa]=0;
1764 if (rlen!=NULL) *rlen=aa;
1765 return translation;
1766 }
1767
1768 void GffObj::printSummary(FILE* fout) {
1769 if (fout==NULL) fout=stdout;
1770 fprintf(fout, "%s\t%c\t%d\t%d\t%4.2f\t%4.1f\n", gffID,
1771 strand, start, end, gscore, (float)qcov/10.0);
1772 }
1773
1774 void GffObj::printGxfLine(FILE* fout, const char* tlabel, const char* gseqname, bool iscds,
1775 uint segstart, uint segend, int exidx, char phase, bool gff3) {
1776 static char scorestr[14];
1777 strcpy(scorestr,".");
1778 GffAttrs* xattrs=NULL;
1779 if (exidx>=0) {
1780 if (exons[exidx]->score) sprintf(scorestr,"%.2f", exons[exidx]->score);
1781 xattrs=exons[exidx]->attrs;
1782 }
1783 if (phase==0 || !iscds) phase='.';
1784 const char* ftype=iscds ? "CDS" : getSubfName();
1785 if (gff3) {
1786 fprintf(fout,
1787 "%s\t%s\t%s\t%d\t%d\t%s\t%c\t%c\tParent=%s",
1788 gseqname, tlabel, ftype, segstart, segend, scorestr, strand,
1789 phase, gffID);
1790 if (xattrs!=NULL) {
1791 for (int i=0;i<xattrs->Count();i++)
1792 fprintf(fout, ";%s=%s",names->attrs.getName(xattrs->Get(i)->attr_id),
1793 xattrs->Get(i)->attr_val);
1794 }
1795 fprintf(fout, "\n");
1796 } //GFF
1797 else {//for GTF -- we print only transcripts
1798 //if (isValidTranscript())
1799 fprintf(fout, "%s\t%s\t%s\t%d\t%d\t%s\t%c\t%c\ttranscript_id \"%s\";",
1800 gseqname, tlabel, ftype, segstart, segend, scorestr, strand, phase, gffID);
1801 //char* geneid=(geneID!=NULL)? geneID : gffID;
1802 if (geneID)
1803 fprintf(fout," gene_id \"%s\";",geneID);
1804 if (gene_name!=NULL) {
1805 //fprintf(fout, " gene_name ");
1806 //if (gene_name[0]=='"') fprintf (fout, "%s;",gene_name);
1807 // else fprintf(fout, "\"%s\";",gene_name);
1808 fprintf(fout," gene_name \"%s\";",gene_name);
1809 }
1810 if (xattrs!=NULL) {
1811 for (int i=0;i<xattrs->Count();i++) {
1812 if (xattrs->Get(i)->attr_val==NULL) continue;
1813 const char* attrname=names->attrs.getName(xattrs->Get(i)->attr_id);
1814 fprintf(fout, " %s ",attrname);
1815 if (xattrs->Get(i)->attr_val[0]=='"')
1816 fprintf(fout, "%s;",xattrs->Get(i)->attr_val);
1817 else fprintf(fout, "\"%s\";",xattrs->Get(i)->attr_val);
1818 }
1819 }
1820 //for GTF, also append the GffObj attributes to each exon line
1821 if ((xattrs=this->attrs)!=NULL) {
1822 for (int i=0;i<xattrs->Count();i++) {
1823 if (xattrs->Get(i)->attr_val==NULL) continue;
1824 const char* attrname=names->attrs.getName(xattrs->Get(i)->attr_id);
1825 fprintf(fout, " %s ",attrname);
1826 if (xattrs->Get(i)->attr_val[0]=='"')
1827 fprintf(fout, "%s;",xattrs->Get(i)->attr_val);
1828 else fprintf(fout, "\"%s\";",xattrs->Get(i)->attr_val);
1829 }
1830 }
1831 fprintf(fout, "\n");
1832 }//GTF
1833 }
1834
1835 void GffObj::printGxf(FILE* fout, GffPrintMode gffp,
1836 const char* tlabel, const char* gfparent) {
1837 static char tmpstr[255];
1838 if (tlabel==NULL) {
1839 tlabel=track_id>=0 ? names->tracks.Get(track_id)->name :
1840 (char*)"gffobj" ;
1841 }
1842 unxcoord();
1843 //if (exons.Count()==0) return;
1844 const char* gseqname=names->gseqs.Get(gseq_id)->name;
1845 bool gff3 = (gffp>=pgffAny);
1846 bool showCDS = (gffp==pgtfAny || gffp==pgtfCDS || gffp==pgffCDS || gffp==pgffAny || gffp==pgffBoth);
1847 bool showExon = (gffp<=pgtfExon || gffp==pgffAny || gffp==pgffExon || gffp==pgffBoth);
1848 if (gff3) {
1849 //print GFF3 mRNA line:
1850 if (gscore>0.0) sprintf(tmpstr,"%.2f", gscore);
1851 else strcpy(tmpstr,".");
1852 uint pstart, pend;
1853 if (gffp==pgffCDS) {
1854 pstart=CDstart;
1855 pend=CDend;
1856 }
1857 else { pstart=start;pend=end; }
1858 //const char* ftype=isTranscript() ? "mRNA" : getFeatureName();
1859 const char* ftype=getFeatureName();
1860 fprintf(fout,
1861 "%s\t%s\t%s\t%d\t%d\t%s\t%c\t.\tID=%s",
1862 gseqname, tlabel, ftype, pstart, pend, tmpstr, strand, gffID);
1863 if (CDstart>0 && !showCDS && !isCDS) fprintf(fout,";CDS=%d-%d",CDstart,CDend);
1864 if (gfparent!=NULL) {
1865 //parent override
1866 fprintf(fout, ";Parent=%s",gfparent);
1867 }
1868 else {
1869 if (parent!=NULL && !parent->isDiscarded())
1870 fprintf(fout, ";Parent=%s",parent->getID());
1871 }
1872 if (geneID!=NULL)
1873 fprintf(fout, ";geneID=%s",geneID);
1874 if (gene_name!=NULL)
1875 fprintf(fout, ";gene_name=%s",gene_name);
1876 if (attrs!=NULL) {
1877 for (int i=0;i<attrs->Count();i++) {
1878 const char* attrname=names->attrs.getName(attrs->Get(i)->attr_id);
1879 fprintf(fout,";%s=%s", attrname,
1880 attrs->Get(i)->attr_val);
1881 }
1882 }
1883 fprintf(fout,"\n");
1884 }// gff3 mRNA line
1885 if (showExon) {
1886 //print exons
1887 if (isCDS && exons.Count()>0 &&
1888 ((strand=='-' && exons.Last()->phase<'0') || (strand=='+' && exons.Last()->phase<'0')))
1889 updateExonPhase();
1890
1891 for (int i=0;i<exons.Count();i++) {
1892 printGxfLine(fout, tlabel, gseqname, isCDS, exons[i]->start, exons[i]->end, i, exons[i]->phase, gff3);
1893 }
1894 }//printing exons
1895 if (showCDS && !isCDS && CDstart>0) {
1896 GArray<GffCDSeg> cds(true,true);
1897 getCDSegs(cds);
1898 for (int i=0;i<cds.Count();i++) {
1899 printGxfLine(fout, tlabel, gseqname, true, cds[i].start, cds[i].end, -1, cds[i].phase, gff3);
1900 }
1901 } //showCDS
1902 }
1903
1904 void GffObj::updateExonPhase() {
1905 if (!isCDS) return;
1906 int cdsacc=0;
1907 if (CDphase=='1' || CDphase=='2') {
1908 cdsacc+= 3-(CDphase-'0');
1909 }
1910 if (strand=='-') { //reverse strand
1911 for (int i=exons.Count()-1;i>=0;i--) {
1912 exons[i]->phase='0'+ (3-cdsacc%3)%3;
1913 cdsacc+=exons[i]->end-exons[i]->start+1;
1914 }
1915 }
1916 else { //forward strand
1917 for (int i=0;i<exons.Count();i++) {
1918 exons[i]->phase='0'+ (3-cdsacc%3)%3;
1919 cdsacc+=exons[i]->end-exons[i]->start+1;
1920 }
1921 }
1922 }
1923
1924
1925 void GffObj::getCDSegs(GArray<GffCDSeg>& cds) {
1926 GffCDSeg cdseg;
1927 int cdsacc=0;
1928 if (CDphase=='1' || CDphase=='2') {
1929 cdsacc+= 3-(CDphase-'0');
1930 }
1931 if (strand=='-') {
1932 for (int x=exons.Count()-1;x>=0;x--) {
1933 uint sgstart=exons[x]->start;
1934 uint sgend=exons[x]->end;
1935 if (CDend<sgstart || CDstart>sgend) continue;
1936 if (CDstart>=sgstart && CDstart<=sgend)
1937 sgstart=CDstart; //cdstart within this segment
1938 if (CDend>=sgstart && CDend<=sgend)
1939 sgend=CDend; //cdend within this segment
1940 cdseg.start=sgstart;
1941 cdseg.end=sgend;
1942 cdseg.exonidx=x;
1943 //cdseg.phase='0'+(cdsacc>0 ? (3-cdsacc%3)%3 : 0);
1944 cdseg.phase='0'+ (3-cdsacc%3)%3;
1945 cdsacc+=sgend-sgstart+1;
1946 cds.Add(cdseg);
1947 } //for each exon
1948 } // - strand
1949 else { // + strand
1950 for (int x=0;x<exons.Count();x++) {
1951 uint sgstart=exons[x]->start;
1952 uint sgend=exons[x]->end;
1953 if (CDend<sgstart || CDstart>sgend) continue;
1954 if (CDstart>=sgstart && CDstart<=sgend)
1955 sgstart=CDstart; //seqstart within this segment
1956 if (CDend>=sgstart && CDend<=sgend)
1957 sgend=CDend; //seqend within this segment
1958 cdseg.start=sgstart;
1959 cdseg.end=sgend;
1960 cdseg.exonidx=x;
1961 //cdseg.phase='0'+(cdsacc>0 ? (3-cdsacc%3)%3 : 0);
1962 cdseg.phase='0' + (3-cdsacc%3)%3 ;
1963 cdsacc+=sgend-sgstart+1;
1964 cds.Add(cdseg);
1965 } //for each exon
1966 } // + strand
1967 }