ViewVC Help
View File | Revision Log | Show Annotations | View Changeset | Root Listing
root/gclib/gclib/gff.cpp
Revision: 153
Committed: Fri Jan 20 22:35:32 2012 UTC (7 years, 9 months ago) by gpertea
File size: 67755 byte(s)
Log Message:
still working on trans-splicing/fusions in gff

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) {
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 -- probably an 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 gffID=Gstrdup(gffline->ID);
819 if (keepAttr) this->parseAttrs(attrs, gffline->info);
820 }
821 }
822 } //subfeature given directly
823 else { //gffline->parents==NULL
824 //create a parent feature in its own right
825 gscore=gffline->score;
826 if (gffline->ID==NULL || gffline->ID[0]==0)
827 GError("Error: no ID found for GFF record start\n");
828 this->hasGffID(true);
829 gffID=Gstrdup(gffline->ID); //there must be an ID here
830 //if (gffline->is_transcript) ftype_id=gff_fid_mRNA;
831 //else
832 ftype_id=names->feats.addName(gffline->ftype);
833 if (gffline->is_transcript)
834 exon_ftype_id=gff_fid_exon;
835
836 if (keepAttr) this->parseAttrs(attrs, gffline->info);
837 }//no parent
838
839 if (gffline->gene_name!=NULL) {
840 gene_name=Gstrdup(gffline->gene_name);
841 }
842 if (gffline->gene_id!=NULL) {
843 geneID=Gstrdup(gffline->gene_id);
844 }
845
846 GSeqStat* gsd=gfrd->gseqstats.AddIfNew(new GSeqStat(gseq_id,names->gseqs.lastNameUsed()),true);
847 uptr=gsd;
848 if (start<gsd->mincoord) gsd->mincoord=start;
849 if (end>gsd->maxcoord) gsd->maxcoord=end;
850 if (this->len()>gsd->maxfeat_len) {
851 gsd->maxfeat_len=this->len();
852 gsd->maxfeat=this;
853 }
854 }
855
856 GffLine* GffReader::nextGffLine() {
857 if (gffline!=NULL) return gffline; //caller should free gffline after processing
858 while (gffline==NULL) {
859 int llen=0;
860 buflen=GFF_LINELEN-1;
861 char* l=fgetline(linebuf, buflen, fh, &fpos, &llen);
862 if (l==NULL) {
863 return NULL; //end of file
864 }
865 int ns=0; //first nonspace position
866 while (l[ns]!=0 && isspace(l[ns])) ns++;
867 if (l[ns]=='#' || llen<10) continue;
868 gffline=new GffLine(this, l);
869 if (gffline->skip) {
870 delete gffline;
871 gffline=NULL;
872 continue;
873 }
874 if (gffline->ID==NULL && gffline->parents==NULL) { //it must have an ID
875 //this might not be needed, already checked in the GffLine constructor
876 if (gff_warns)
877 GMessage("Warning: malformed GFF line, no parent or record Id (kipping\n");
878 delete gffline;
879 gffline=NULL;
880 //continue;
881 }
882 }
883 return gffline;
884 }
885
886
887 char* GffReader::gfoBuildId(const char* id, const char* ctg) {
888 //caller must free the returned pointer
889 char* buf=NULL;
890 int idlen=strlen(id);
891 GMALLOC(buf, idlen+strlen(ctg)+2);
892 strcpy(buf, id);
893 buf[idlen]='~';
894 strcpy(buf+idlen+1, ctg);
895 return buf;
896 }
897 /*
898 void GffReader::gfoRemove(const char* id, const char* ctg) {
899 char* buf=gfoBuildId(id,ctg);
900 phash.Remove(buf);
901 GFREE(buf);
902 }
903 */
904 //Warning: if gflst gets altered, idx becomes obsolete
905 GfoHolder* GffReader::gfoAdd(GffObj* gfo, int idx) {
906 GVec<GfoHolder>* glst=new GVec<GfoHolder>(1);
907 GfoHolder gh(gfo,idx);
908 int i=glst->Add(gh);
909 phash.Add(gfo->gffID, glst);
910 return &(glst->Get(i));
911 }
912
913 GfoHolder* GffReader::gfoAdd(GVec<GfoHolder>& glst, GffObj* gfo, int idx) {
914 GfoHolder gh(gfo,idx);
915 int i=glst.Add(gh);
916 return &(glst[i]);
917 }
918
919 GfoHolder* GffReader::gfoFind(const char* id, const char* ctg, char strand, uint start, GVec<GfoHolder>** glst) {
920 GVec<GfoHolder>* gl=phash.Find(id);
921 GfoHolder* gh=NULL;
922 if (gl) {
923 for (int i=0;i<gl->Count();i++) {
924 GfoHolder& gfo = gl->Get(i);
925 if (ctg!=NULL && strcmp(ctg,gfo.gffobj->getGSeqName())!=0)
926 continue;
927 if (strand && strand != gfo.gffobj->strand)
928 continue;
929 if (start>0 && abs((int)start-(int)gfo.gffobj->start)>GFF_MAX_LOCUS)
930 continue;
931 //must be the same transcript, according to given comparison criteria
932 gh=&gfo;
933 break;
934 }
935 }
936 if (glst) *glst=gl;
937 return gh;
938 }
939
940 GfoHolder* GffReader::replaceGffRec(GffLine* gffline, bool keepAttr, bool noExonAttr, int replaceidx) {
941 GffObj* newgfo=new GffObj(this, gffline, keepAttr, noExonAttr);
942 GfoHolder* r=NULL;
943 if (replaceidx>=0) {
944 gflst.Put(replaceidx,newgfo);
945 r=gfoAdd(newgfo, replaceidx);
946 }
947 else {
948 int gfoidx=gflst.Add(newgfo);
949 r=gfoAdd(newgfo, gfoidx);
950 }
951 /*
952 if (gff_warns) {
953 int* pcount=tids.Find(newgfo->gffID);
954 if (pcount!=NULL) {
955 if (gff_warns) GMessage("Warning: duplicate GFF ID: %s\n", newgfo->gffID);
956 (*pcount)++;
957 }
958 else {
959 tids.Add(newgfo->gffID,new int(1));
960 }
961 }
962 */
963 return r;
964 }
965
966 GfoHolder* GffReader::updateParent(GfoHolder* newgfh, GffObj* parent) {
967 //assert(parent);
968 //assert(newgfo);
969 parent->children.Add(newgfh->gffobj);
970 if (newgfh->gffobj->parent==NULL) newgfh->gffobj->parent=parent;
971 newgfh->gffobj->setLevel(parent->getLevel()+1);
972 if (parent->isGene()) {
973 if (parent->gene_name!=NULL && newgfh->gffobj->gene_name==NULL)
974 newgfh->gffobj->gene_name=Gstrdup(parent->gene_name);
975 if (parent->geneID!=NULL && newgfh->gffobj->geneID==NULL)
976 newgfh->gffobj->geneID=Gstrdup(parent->geneID);
977 }
978
979 return newgfh;
980 }
981
982 GfoHolder* GffReader::newGffRec(GffLine* gffline, bool keepAttr, bool noExonAttr,
983 GffObj* parent, GffExon* pexon, GVec<GfoHolder>* glst) {
984 GffObj* newgfo=new GffObj(this, gffline, keepAttr, noExonAttr);
985 GfoHolder* r=NULL;
986 int gfoidx=gflst.Add(newgfo);
987 r=(glst) ? gfoAdd(*glst, newgfo, gfoidx) :gfoAdd(newgfo, gfoidx);
988 if (parent!=NULL) {
989 updateParent(r, parent);
990 if (pexon!=NULL) parent->removeExon(pexon);
991 }
992 /*
993 if (gff_warns) {
994 int* pcount=tids.Find(newgfo->gffID);
995 if (pcount!=NULL) {
996 if (gff_warns) GMessage("Warning: duplicate GFF ID: %s\n", newgfo->gffID);
997 (*pcount)++;
998 }
999 else {
1000 tids.Add(newgfo->gffID,new int(1));
1001 }
1002 }
1003 */
1004 return r;
1005 }
1006
1007 GfoHolder* GffReader::updateGffRec(GfoHolder* prevgfo, GffLine* gffline,
1008 bool keepAttr) {
1009 if (prevgfo==NULL) return NULL;
1010 //prevgfo->gffobj->createdByExon(false);
1011 prevgfo->gffobj->ftype_id=prevgfo->gffobj->names->feats.addName(gffline->ftype);
1012 prevgfo->gffobj->start=gffline->fstart;
1013 prevgfo->gffobj->end=gffline->fend;
1014 prevgfo->gffobj->isGene(gffline->is_gene);
1015 prevgfo->gffobj->isTranscript(gffline->is_transcript || gffline->exontype!=0);
1016 prevgfo->gffobj->hasGffID(gffline->ID!=NULL);
1017 if (keepAttr) {
1018 if (prevgfo->gffobj->attrs!=NULL) prevgfo->gffobj->attrs->Clear();
1019 prevgfo->gffobj->parseAttrs(prevgfo->gffobj->attrs, gffline->info);
1020 }
1021 return prevgfo;
1022 }
1023
1024
1025 bool GffReader::addExonFeature(GfoHolder* prevgfo, GffLine* gffline, GHash<CNonExon>& pex, bool noExonAttr) {
1026 bool r=true;
1027 if (gffline->strand!=prevgfo->gffobj->strand) {
1028 //TODO: add support for trans-splicing and even inter-chromosomal fusions
1029 if (prevgfo->gffobj->strand=='.') {
1030 prevgfo->gffobj->strand=gffline->strand;
1031 }
1032 else {
1033 GMessage("GFF Error at %s (%c): exon %d-%d (%c) found on different strand; discarded.\n",
1034 prevgfo->gffobj->gffID, prevgfo->gffobj->strand,
1035 gffline->fstart, gffline->fend, gffline->strand, prevgfo->gffobj->getGSeqName());
1036 //r=false;
1037 return true; //FIXME: split trans-spliced mRNAs by strand
1038 }
1039 }
1040 int gdist=(gffline->fstart>prevgfo->gffobj->end) ? gffline->fstart-prevgfo->gffobj->end :
1041 ((gffline->fend<prevgfo->gffobj->start)? prevgfo->gffobj->start-gffline->fend :
1042 0 );
1043 if (gdist>(int)GFF_MAX_LOCUS) { //too far apart, most likely this is a duplicate ID
1044 GMessage("Error: duplicate GFF ID '%s' (or exons too far apart)!\n",prevgfo->gffobj->gffID);
1045 //validation_errors = true;
1046 r=false;
1047 if (!gff_warns) exit(1);
1048 }
1049 int eidx=prevgfo->gffobj->addExon(this, gffline, !noExonAttr, noExonAttr);
1050 if (eidx>=0 && gffline->ID!=NULL && gffline->exontype==0)
1051 subfPoolAdd(pex, prevgfo);
1052 return r;
1053 }
1054
1055 CNonExon* GffReader::subfPoolCheck(GffLine* gffline, GHash<CNonExon>& pex, char*& subp_name) {
1056 CNonExon* subp=NULL;
1057 subp_name=NULL;
1058 for (int i=0;i<gffline->num_parents;i++) {
1059 if (transcriptsOnly && discarded_ids.Find(gffline->parents[i])!=NULL)
1060 continue;
1061 subp_name=gfoBuildId(gffline->parents[i], gffline->gseqname); //e.g. mRNA name
1062 subp=pex.Find(subp_name);
1063 if (subp!=NULL)
1064 return subp;
1065 GFREE(subp_name);
1066 }
1067 return NULL;
1068 }
1069
1070 void GffReader::subfPoolAdd(GHash<CNonExon>& pex, GfoHolder* newgfo) {
1071 //this might become a parent feature later
1072 if (newgfo->gffobj->exons.Count()>0) {
1073 char* xbuf=gfoBuildId(gffline->ID, gffline->gseqname);
1074 pex.Add(xbuf, new CNonExon(newgfo->idx, newgfo->gffobj,
1075 newgfo->gffobj->exons[0], gffline));
1076 GFREE(xbuf);
1077 }
1078 }
1079
1080 GfoHolder* GffReader::promoteFeature(CNonExon* subp, char*& subp_name, GHash<CNonExon>& pex,
1081 bool keepAttr, bool noExonAttr) {
1082 GffObj* prevp=subp->parent; //grandparent of gffline (e.g. gene)
1083 if (prevp!=gflst[subp->idx])
1084 GError("Error promoting subfeature %s, gflst index mismatch?!\n", subp->gffline->ID);
1085 subp->gffline->discardParent();
1086 GfoHolder* gfoh=newGffRec(subp->gffline, keepAttr, noExonAttr, prevp, subp->exon);
1087 pex.Remove(subp_name); //no longer a potential parent, moved it to phash already
1088 prevp->promotedChildren(true);
1089 return gfoh; //returns the holder of newly promoted feature
1090 }
1091
1092 //have to parse the whole file because exons can be scattered all over
1093 //trans-splicing and fusions are only accepted in proper GFF3 format, with a single parent feature ID entry
1094 void GffReader::readAll(bool keepAttr, bool mergeCloseExons, bool noExonAttr) {
1095 bool validation_errors = false;
1096 //loc_debug=false;
1097 GHash<CNonExon> pex; //keep track of any "exon"-like features that have an ID
1098 //and thus could become promoted to parent features
1099 while (nextGffLine()!=NULL) {
1100 //seen this gff ID before?
1101 GfoHolder* prevseen=NULL;
1102 if (gffline->ID && gffline->exontype==0) //GFF3 parent-like feature (mRNA, gene, etc.)
1103 //look for same ID on the same chromosome
1104 prevseen=gfoFind(gffline->ID, gffline->gseqname);
1105 if (prevseen!=NULL) {
1106 //found same ID/chromosome combo
1107 if (prevseen->gffobj->createdByExon() &&
1108 prevseen->gffobj->start>=gffline->fstart && prevseen->gffobj->end<=gffline->fend) {
1109 //an exon of this ID was given before
1110 //this line has the main attributes for this ID
1111 updateGffRec(prevseen, gffline, keepAttr);
1112 }
1113 else {
1114 GMessage("Error: duplicate GFF ID '%s' encountered!\n",gffline->ID);
1115 validation_errors = true;
1116 if (gff_warns) {
1117 delete gffline; gffline=NULL; continue;
1118 }
1119 else exit(1);
1120 }
1121 }
1122 if (gffline->parents==NULL) {//start GFF3-like record with no parent (mRNA, gene)
1123 if (!prevseen) newGffRec(gffline, keepAttr, noExonAttr);
1124 }
1125 else { //--- it's a parented feature (exon/CDS, but might still be a mRNA)
1126 bool found_parent=false;
1127 GfoHolder* newgfo=prevseen;
1128 GVec<GfoHolder>* newgflst=NULL;
1129 for (int i=0;i<gffline->num_parents;i++) {
1130 if (transcriptsOnly && discarded_ids.Find(gffline->parents[i])!=NULL)
1131 continue; //skipping discarded parent feature
1132 GfoHolder* parentgfo=gfoFind(gffline->parents[i], gffline->gseqname, gffline->strand, gffline->fstart, &newgflst);
1133 if (parentgfo!=NULL) { //parent GffObj parsed earlier
1134 found_parent=true;
1135 if (parentgfo->gffobj->isGene() && gffline->is_transcript
1136 && gffline->exontype==0) {
1137 //not an exon, but a transcript parented by a gene
1138 if (newgfo) {
1139 updateParent(newgfo, parentgfo->gffobj);
1140 }
1141 else {
1142 newgfo=newGffRec(gffline, keepAttr, noExonAttr, parentgfo->gffobj);
1143 }
1144 }
1145 else { //potential exon subfeature
1146 if (!addExonFeature(parentgfo, gffline, pex, noExonAttr))
1147 validation_errors=true;
1148 }
1149 }
1150 } //for each parsed parent Id
1151 if (!found_parent) { //new GTF-like record starting here with a subfeature directly
1152 //or it could be some chado GFF3 barf with exons declared BEFORE their parent :(
1153 //check if this feature isn't parented by a previously stored "exon" subfeature
1154 char* subp_name=NULL;
1155 CNonExon* subp=subfPoolCheck(gffline, pex, subp_name);
1156 if (subp!=NULL) { //found a subfeature that is the parent of this gffline
1157 //promote that subfeature to a full GffObj
1158 GfoHolder* gfoh=promoteFeature(subp, subp_name, pex, keepAttr, noExonAttr);
1159 //add current gffline as an exon of the newly promoted subfeature
1160 if (!addExonFeature(gfoh, gffline, pex, noExonAttr))
1161 validation_errors=true;
1162 }
1163 else { //no parent seen before, create one directly with this exon
1164 //loc_debug=true;
1165 GfoHolder* newgfo=prevseen ? prevseen : newGffRec(gffline, keepAttr, noExonAttr, NULL, NULL, newgflst);
1166 if (gffline->ID!=NULL && gffline->exontype==0)
1167 subfPoolAdd(pex, newgfo);
1168 //even those with errors will be added here!
1169 }
1170 GFREE(subp_name);
1171 } //no previous parent found
1172 } //parented feature
1173 //--
1174 delete gffline;
1175 gffline=NULL;
1176 }//while gff lines
1177 gflst.finalize(this, mergeCloseExons, keepAttr, noExonAttr); //force sorting by locus if so constructed
1178 // all gff records are now loaded in GList gflst
1179 // so we can free the hash
1180 phash.Clear();
1181 //tids.Clear();
1182 if (validation_errors) {
1183 exit(1);
1184 }
1185 }
1186
1187 GffObj* GffObj::finalize(GffReader* gfr, bool mergeCloseExons, bool keepAttrs, bool noExonAttr) {
1188 //merge
1189 //always merge adjacent or overlapping segments
1190 //but if mergeCloseExons then merge even when distance is up to 5 bases
1191 udata=0;
1192 uptr=NULL;
1193 if (gfr->transcriptsOnly && !(isTranscript() || (isGene() && children.Count()==0))) {
1194 isDiscarded(true);
1195 }
1196 if (ftype_id==gff_fid_transcript && CDstart>0) {
1197 ftype_id=gff_fid_mRNA;
1198 //exon_ftype_id=gff_fid_exon;
1199 }
1200 //if (ftype_id==gff_fid_mRNA || exon_ftype_id==gff_fid_exon || mergeCloseExons) {
1201 if (isTranscript() || exon_ftype_id==gff_fid_exon || mergeCloseExons) {
1202 int mindist=mergeCloseExons ? 5:1;
1203 for (int i=0;i<exons.Count()-1;i++) {
1204 int ni=i+1;
1205 uint mend=exons[i]->end;
1206 while (ni<exons.Count()) {
1207 int dist=(int)(exons[ni]->start-mend);
1208 if (dist>mindist) break; //no merging with next segment
1209 if (gfr!=NULL && gfr->gff_warns && dist!=0 && (exons[ni]->exontype!=exgffUTR && exons[i]->exontype!=exgffUTR)) {
1210 GMessage("GFF warning: merging adjacent/overlapping segments of %s on %s (%d-%d, %d-%d)\n",
1211 gffID, getGSeqName(), exons[i]->start, exons[i]->end,exons[ni]->start, exons[ni]->end);
1212 }
1213 mend=exons[ni]->end;
1214 covlen-=exons[i]->len();
1215 exons[i]->end=mend;
1216 covlen+=exons[i]->len();
1217 covlen-=exons[ni]->len();
1218 if (exons[ni]->attrs!=NULL && (exons[i]->attrs==NULL ||
1219 exons[i]->attrs->Count()<exons[ni]->attrs->Count())) {
1220 //use the other exon attributes, if more
1221 delete(exons[i]->attrs);
1222 exons[i]->attrs=exons[ni]->attrs;
1223 exons[ni]->attrs=NULL;
1224 }
1225 exons.Delete(ni);
1226 } //check for merge with next exon
1227 } //for each exon
1228 }
1229 //attribute reduction for GTF records
1230 if (keepAttrs && !noExonAttr && !hasGffID()
1231 && exons.Count()>0 && exons[0]->attrs!=NULL) {
1232 bool attrs_discarded=false;
1233 for (int a=0;a<exons[0]->attrs->Count();a++) {
1234 int attr_name_id=exons[0]->attrs->Get(a)->attr_id;
1235 char* attr_name=names->attrs.getName(attr_name_id);
1236 char* attr_val =exons[0]->attrs->Get(a)->attr_val;
1237 bool sameExonAttr=true;
1238 for (int i=1;i<exons.Count();i++) {
1239 char* ov=exons[i]->getAttr(attr_name_id);
1240 if (ov==NULL || (strcmp(ov,attr_val)!=0)) {
1241 sameExonAttr=false;
1242 break;
1243 }
1244 }
1245 if (sameExonAttr) {
1246 //delete this attribute from exons level
1247 attrs_discarded=true;
1248 this->addAttr(attr_name, attr_val);
1249 for (int i=1;i<exons.Count();i++) {
1250 removeExonAttr(*(exons[i]), attr_name_id);
1251 }
1252 exons[0]->attrs->freeItem(a);
1253 }
1254 }
1255 if (attrs_discarded) exons[0]->attrs->Pack();
1256 }
1257 return this;
1258 }
1259
1260 void GffObj::parseAttrs(GffAttrs*& atrlist, char* info, bool isExon) {
1261 if (names==NULL)
1262 GError(ERR_NULL_GFNAMES, "parseAttrs()");
1263 if (atrlist==NULL)
1264 atrlist=new GffAttrs();
1265 char* endinfo=info+strlen(info);
1266 char* start=info;
1267 char* pch=start;
1268 while (start<endinfo) {
1269 //skip spaces
1270 while (*start==' ' && start<endinfo) start++;
1271 pch=strchr(start, ';');
1272 if (pch==NULL) pch=endinfo;
1273 else {
1274 *pch='\0';
1275 pch++;
1276 }
1277 char* ech=strchr(start,'=');
1278 if (ech!=NULL) { // attr=value format found
1279 *ech='\0';
1280 //if (noExonAttr && (strcmp(start, "exon_number")==0 || strcmp(start, "exon")==0)) { start=pch; continue; }
1281 if (strcmp(start, "exon_number")==0 || strcmp(start, "exon")==0 ||
1282 strcmp(start, "exon_id")==0)
1283 { start=pch; continue; }
1284 ech++;
1285 while (*ech==' ' && ech<endinfo) ech++;//skip extra spaces after the '='
1286 //atrlist->Add(new GffAttr(names->attrs.addName(start),ech));
1287 //make sure we don't add the same attribute more than once
1288 if (isExon && (strcmp(start, "protein_id")==0)) {
1289 //Ensembl special case
1290 this->addAttr(start, ech);
1291 start=pch;
1292 continue;
1293 }
1294 atrlist->add_or_update(names, start, ech);
1295 }
1296 /*
1297 else { //not an attr=value format
1298 atrlist->Add(new GffAttr(names->attrs.addName(start),"1"));
1299 }
1300 */
1301 start=pch;
1302 }
1303 if (atrlist->Count()==0) { delete atrlist; atrlist=NULL; }
1304 }
1305
1306 void GffObj::addAttr(const char* attrname, const char* attrvalue) {
1307 if (this->attrs==NULL)
1308 this->attrs=new GffAttrs();
1309 //this->attrs->Add(new GffAttr(names->attrs.addName(attrname),attrvalue));
1310 this->attrs->add_or_update(names, attrname, attrvalue);
1311 }
1312
1313
1314 void GffObj::setFeatureName(const char* feature) {
1315 //change the feature name/type for a transcript
1316 int fid=names->feats.addName(feature);
1317 if (monoFeature() && exons.Count()>0)
1318 this->exon_ftype_id=fid;
1319 this->ftype_id=fid;
1320 }
1321
1322 void GffObj::setRefName(const char* newname) {
1323 //change the feature name/type for a transcript
1324 int rid=names->gseqs.addName(newname);
1325 this->gseq_id=rid;
1326 }
1327
1328
1329
1330 int GffObj::removeAttr(const char* attrname, const char* attrval) {
1331 if (this->attrs==NULL || attrname==NULL || attrname[0]==0) return 0;
1332 int aid=this->names->attrs.getId(attrname);
1333 if (aid<0) return 0;
1334 int delcount=0; //could be more than one ?
1335 for (int i=0;i<this->attrs->Count();i++) {
1336 if (aid==this->attrs->Get(i)->attr_id) {
1337 if (attrval==NULL ||
1338 strcmp(attrval, this->attrs->Get(i)->attr_val)==0) {
1339 delcount++;
1340 this->attrs->freeItem(i);
1341 }
1342 }
1343 }
1344 if (delcount>0) this->attrs->Pack();
1345 return delcount;
1346 }
1347
1348 int GffObj::removeAttr(int aid, const char* attrval) {
1349 if (this->attrs==NULL || aid<0) return 0;
1350 int delcount=0; //could be more than one ?
1351 for (int i=0;i<this->attrs->Count();i++) {
1352 if (aid==this->attrs->Get(i)->attr_id) {
1353 if (attrval==NULL ||
1354 strcmp(attrval, this->attrs->Get(i)->attr_val)==0) {
1355 delcount++;
1356 this->attrs->freeItem(i);
1357 }
1358 }
1359 }
1360 if (delcount>0) this->attrs->Pack();
1361 return delcount;
1362 }
1363
1364
1365 int GffObj::removeExonAttr(GffExon& exon, const char* attrname, const char* attrval) {
1366 if (exon.attrs==NULL || attrname==NULL || attrname[0]==0) return 0;
1367 int aid=this->names->attrs.getId(attrname);
1368 if (aid<0) return 0;
1369 int delcount=0; //could be more than one
1370 for (int i=0;i<exon.attrs->Count();i++) {
1371 if (aid==exon.attrs->Get(i)->attr_id) {
1372 if (attrval==NULL ||
1373 strcmp(attrval, exon.attrs->Get(i)->attr_val)==0) {
1374 delcount++;
1375 exon.attrs->freeItem(i);
1376 }
1377 }
1378 }
1379 if (delcount>0) exon.attrs->Pack();
1380 return delcount;
1381 }
1382
1383 int GffObj::removeExonAttr(GffExon& exon, int aid, const char* attrval) {
1384 if (exon.attrs==NULL || aid<0) return 0;
1385 int delcount=0; //could be more than one
1386 for (int i=0;i<exon.attrs->Count();i++) {
1387 if (aid==exon.attrs->Get(i)->attr_id) {
1388 if (attrval==NULL ||
1389 strcmp(attrval, exon.attrs->Get(i)->attr_val)==0) {
1390 delcount++;
1391 exon.attrs->freeItem(i);
1392 }
1393 }
1394 }
1395 if (delcount>0) exon.attrs->Pack();
1396 return delcount;
1397 }
1398
1399
1400 void GffObj::getCDS_ends(uint& cds_start, uint& cds_end) {
1401 cds_start=0;
1402 cds_end=0;
1403 if (CDstart==0 || CDend==0) return; //no CDS info
1404 int cdsadj=0;
1405 if (CDphase=='1' || CDphase=='2') {
1406 cdsadj=CDphase-'0';
1407 }
1408 cds_start=CDstart;
1409 cds_end=CDend;
1410 if (strand=='-') cds_end-=cdsadj;
1411 else cds_start+=cdsadj;
1412 }
1413
1414 void GffObj::mRNA_CDS_coords(uint& cds_mstart, uint& cds_mend) {
1415 //sets cds_start and cds_end to the CDS start,end coordinates on the spliced mRNA transcript
1416 cds_mstart=0;
1417 cds_mend=0;
1418 if (CDstart==0 || CDend==0) return; //no CDS info
1419 //restore normal coordinates, just in case
1420 unxcoord();
1421 int cdsadj=0;
1422 if (CDphase=='1' || CDphase=='2') {
1423 cdsadj=CDphase-'0';
1424 }
1425 /*
1426 uint seqstart=CDstart;
1427 uint seqend=CDend;
1428 */
1429 uint seqstart=exons.First()->start;
1430 uint seqend=exons.Last()->end;
1431 int s=0; //resulting nucleotide counter
1432 if (strand=='-') {
1433 for (int x=exons.Count()-1;x>=0;x--) {
1434 uint sgstart=exons[x]->start;
1435 uint sgend=exons[x]->end;
1436 if (seqend<sgstart || seqstart>sgend) continue;
1437 if (seqstart>=sgstart && seqstart<=sgend)
1438 sgstart=seqstart; //seqstart within this segment
1439 if (seqend>=sgstart && seqend<=sgend)
1440 sgend=seqend; //seqend within this segment
1441 s+=(int)(sgend-sgstart)+1;
1442 if (CDstart>=sgstart && CDstart<=sgend) {
1443 //CDstart in this segment
1444 //and we are getting the whole transcript
1445 cds_mend=s-(int)(CDstart-sgstart);
1446 }
1447 if (CDend>=sgstart && CDend<=sgend) {
1448 //CDstart in this segment
1449 //and we are getting the whole transcript
1450 cds_mstart=s-(int)(CDend-cdsadj-sgstart);
1451 }
1452 } //for each exon
1453 } // - strand
1454 else { // + strand
1455 for (int x=0;x<exons.Count();x++) {
1456 uint sgstart=exons[x]->start;
1457 uint sgend=exons[x]->end;
1458 if (seqend<sgstart || seqstart>sgend) continue;
1459 if (seqstart>=sgstart && seqstart<=sgend)
1460 sgstart=seqstart; //seqstart within this segment
1461 if (seqend>=sgstart && seqend<=sgend)
1462 sgend=seqend; //seqend within this segment
1463 s+=(int)(sgend-sgstart)+1;
1464 /* for (uint i=sgstart;i<=sgend;i++) {
1465 spliced[s]=gsubseq[i-gstart];
1466 s++;
1467 }//for each nt
1468 */
1469 if (CDstart>=sgstart && CDstart<=sgend) {
1470 //CDstart in this segment
1471 cds_mstart=s-(int)(sgend-CDstart-cdsadj);
1472 }
1473 if (CDend>=sgstart && CDend<=sgend) {
1474 //CDend in this segment
1475 cds_mend=s-(int)(sgend-CDend);
1476 }
1477 } //for each exon
1478 } // + strand
1479 //spliced[s]=0;
1480 //if (rlen!=NULL) *rlen=s;
1481 //return spliced;
1482 }
1483
1484 char* GffObj::getUnspliced(GFaSeqGet* faseq, int* rlen, GList<GSeg>* seglst)
1485 {
1486 if (faseq==NULL) { GMessage("Warning: getUnspliced(NULL,.. ) called!\n");
1487 return NULL;
1488 }
1489 //restore normal coordinates:
1490 unxcoord();
1491 if (exons.Count()==0) return NULL;
1492 int fspan=end-start+1;
1493 const char* gsubseq=faseq->subseq(start, fspan);
1494 if (gsubseq==NULL) {
1495 GError("Error getting subseq for %s (%d..%d)!\n", gffID, start, end);
1496 }
1497 char* unspliced=NULL;
1498
1499 int seqstart=exons.First()->start;
1500 int seqend=exons.Last()->end;
1501
1502 int unsplicedlen = 0;
1503
1504 unsplicedlen += seqend - seqstart + 1;
1505
1506 GMALLOC(unspliced, unsplicedlen+1); //allocate more here
1507 //uint seqstart, seqend;
1508
1509 int s = 0; //resulting nucleotide counter
1510 if (strand=='-')
1511 {
1512 if (seglst!=NULL)
1513 seglst->Add(new GSeg(s+1,s+1+seqend-seqstart));
1514 for (int i=seqend;i>=seqstart;i--)
1515 {
1516 unspliced[s] = ntComplement(gsubseq[i-start]);
1517 s++;
1518 }//for each nt
1519 } // - strand
1520 else
1521 { // + strand
1522 if (seglst!=NULL)
1523 seglst->Add(new GSeg(s+1,s+1+seqend-seqstart));
1524 for (int i=seqstart;i<=seqend;i++)
1525 {
1526 unspliced[s]=gsubseq[i-start];
1527 s++;
1528 }//for each nt
1529 } // + strand
1530 //assert(s <= unsplicedlen);
1531 unspliced[s]=0;
1532 if (rlen!=NULL) *rlen=s;
1533 return unspliced;
1534 }
1535
1536 char* GffObj::getSpliced(GFaSeqGet* faseq, bool CDSonly, int* rlen, uint* cds_start, uint* cds_end,
1537 GList<GSeg>* seglst) {
1538 if (CDSonly && CDstart==0) return NULL;
1539 if (faseq==NULL) { GMessage("Warning: getSpliced(NULL,.. ) called!\n");
1540 return NULL;
1541 }
1542 //restore normal coordinates:
1543 unxcoord();
1544 if (exons.Count()==0) return NULL;
1545 int fspan=end-start+1;
1546 const char* gsubseq=faseq->subseq(start, fspan);
1547 if (gsubseq==NULL) {
1548 GError("Error getting subseq for %s (%d..%d)!\n", gffID, start, end);
1549 }
1550 if (fspan<(int)(end-start+1)) { //special case: stop coordinate was extended past the gseq length, must adjust
1551 int endadj=end-start+1-fspan;
1552 uint prevend=end;
1553 end-=endadj;
1554 if (CDend>end) CDend=end;
1555 if (exons.Last()->end>end) {
1556 exons.Last()->end=end; //this could get us into trouble if exon start is also > end
1557 if (exons.Last()->start>exons.Last()->end) {
1558 GError("GffObj::getSpliced() error: improper genomic coordinate %d on %s for %s\n",
1559 prevend,getGSeqName(), getID());
1560 }
1561 covlen-=endadj;
1562 }
1563 }
1564 char* spliced=NULL;
1565 GMALLOC(spliced, covlen+1); //allocate more here
1566 uint seqstart, seqend;
1567 int cdsadj=0;
1568 if (CDphase=='1' || CDphase=='2') {
1569 cdsadj=CDphase-'0';
1570 }
1571 if (CDSonly) {
1572 seqstart=CDstart;
1573 seqend=CDend;
1574 if (strand=='-') seqend-=cdsadj;
1575 else seqstart+=cdsadj;
1576 }
1577 else {
1578 seqstart=exons.First()->start;
1579 seqend=exons.Last()->end;
1580 }
1581 int s=0; //resulting nucleotide counter
1582 if (strand=='-') {
1583 for (int x=exons.Count()-1;x>=0;x--) {
1584 uint sgstart=exons[x]->start;
1585 uint sgend=exons[x]->end;
1586 if (seqend<sgstart || seqstart>sgend) continue;
1587 if (seqstart>=sgstart && seqstart<=sgend)
1588 sgstart=seqstart; //seqstart within this segment
1589 if (seqend>=sgstart && seqend<=sgend)
1590 sgend=seqend; //seqend within this segment
1591 if (seglst!=NULL)
1592 seglst->Add(new GSeg(s+1,s+1+sgend-sgstart));
1593 for (uint i=sgend;i>=sgstart;i--) {
1594 spliced[s] = ntComplement(gsubseq[i-start]);
1595 s++;
1596 }//for each nt
1597
1598 if (!CDSonly && cds_start!=NULL && CDstart>0) {
1599 if (CDstart>=sgstart && CDstart<=sgend) {
1600 //CDstart in this segment
1601 //and we are getting the whole transcript
1602 *cds_end=s-(CDstart-sgstart);
1603 }
1604 if (CDend>=sgstart && CDend<=sgend) {
1605 //CDstart in this segment
1606 //and we are getting the whole transcript
1607 *cds_start=s-(CDend-cdsadj-sgstart);
1608 }
1609 }//update local CDS coordinates
1610 } //for each exon
1611 } // - strand
1612 else { // + strand
1613 for (int x=0;x<exons.Count();x++) {
1614 uint sgstart=exons[x]->start;
1615 uint sgend=exons[x]->end;
1616 if (seqend<sgstart || seqstart>sgend) continue;
1617 if (seqstart>=sgstart && seqstart<=sgend)
1618 sgstart=seqstart; //seqstart within this segment
1619 if (seqend>=sgstart && seqend<=sgend)
1620 sgend=seqend; //seqend within this segment
1621 if (seglst!=NULL)
1622 seglst->Add(new GSeg(s+1,s+1+sgend-sgstart));
1623 for (uint i=sgstart;i<=sgend;i++) {
1624 spliced[s]=gsubseq[i-start];
1625 s++;
1626 }//for each nt
1627 if (!CDSonly && cds_start!=NULL && CDstart>0) {
1628 if (CDstart>=sgstart && CDstart<=sgend) {
1629 //CDstart in this segment
1630 //and we are getting the whole transcript
1631 *cds_start=s-(sgend-CDstart-cdsadj);
1632 }
1633 if (CDend>=sgstart && CDend<=sgend) {
1634 //CDstart in this segment
1635 //and we are getting the whole transcript
1636 *cds_end=s-(sgend-CDend);
1637 }
1638 }//update local CDS coordinates
1639 } //for each exon
1640 } // + strand
1641 spliced[s]=0;
1642 if (rlen!=NULL) *rlen=s;
1643 return spliced;
1644 }
1645
1646 char* GffObj::getSplicedTr(GFaSeqGet* faseq, bool CDSonly, int* rlen) {
1647 if (CDSonly && CDstart==0) return NULL;
1648 //restore normal coordinates:
1649 unxcoord();
1650 if (exons.Count()==0) return NULL;
1651 int fspan=end-start+1;
1652 const char* gsubseq=faseq->subseq(start, fspan);
1653 if (gsubseq==NULL) {
1654 GError("Error getting subseq for %s (%d..%d)!\n", gffID, start, end);
1655 }
1656
1657 char* translation=NULL;
1658 GMALLOC(translation, (int)(covlen/3)+1);
1659 uint seqstart, seqend;
1660 int cdsadj=0;
1661 if (CDphase=='1' || CDphase=='2') {
1662 cdsadj=CDphase-'0';
1663 }
1664 if (CDSonly) {
1665 seqstart=CDstart;
1666 seqend=CDend;
1667 if (strand=='-') seqend-=cdsadj;
1668 else seqstart+=cdsadj;
1669 }
1670 else {
1671 seqstart=exons.First()->start;
1672 seqend=exons.Last()->end;
1673 }
1674 Codon codon;
1675 int nt=0; //codon nucleotide counter (0..2)
1676 int aa=0; //aminoacid count
1677 if (strand=='-') {
1678 for (int x=exons.Count()-1;x>=0;x--) {
1679 uint sgstart=exons[x]->start;
1680 uint sgend=exons[x]->end;
1681 if (seqend<sgstart || seqstart>sgend) continue;
1682 if (seqstart>=sgstart && seqstart<=sgend)
1683 sgstart=seqstart; //seqstart within this segment
1684 if (seqend>=sgstart && seqend<=sgend) {
1685 sgend=seqend; //seqend within this segment
1686 }
1687 for (uint i=sgend;i>=sgstart;i--) {
1688 codon.nuc[nt]=ntComplement(gsubseq[i-start]);
1689 nt++;
1690 if (nt==3) {
1691 nt=0;
1692 translation[aa]=codon.translate();
1693 aa++;
1694 }
1695 }//for each nt
1696 } //for each exon
1697 } // - strand
1698 else { // + strand
1699 for (int x=0;x<exons.Count();x++) {
1700 uint sgstart=exons[x]->start;
1701 uint sgend=exons[x]->end;
1702 if (seqend<sgstart || seqstart>sgend) continue;
1703 if (seqstart>=sgstart && seqstart<=sgend)
1704 sgstart=seqstart; //seqstart within this segment
1705 if (seqend>=sgstart && seqend<=sgend)
1706 sgend=seqend; //seqend within this segment
1707 for (uint i=sgstart;i<=sgend;i++) {
1708 codon.nuc[nt]=gsubseq[i-start];
1709 nt++;
1710 if (nt==3) {
1711 nt=0;
1712 translation[aa]=codon.translate();
1713 aa++;
1714 }
1715 }//for each nt
1716 } //for each exon
1717 } // + strand
1718 translation[aa]=0;
1719 if (rlen!=NULL) *rlen=aa;
1720 return translation;
1721 }
1722
1723 void GffObj::printSummary(FILE* fout) {
1724 if (fout==NULL) fout=stdout;
1725 fprintf(fout, "%s\t%c\t%d\t%d\t%4.2f\t%4.1f\n", gffID,
1726 strand, start, end, gscore, (float)qcov/10.0);
1727 }
1728
1729 void GffObj::printGxfLine(FILE* fout, const char* tlabel, const char* gseqname, bool iscds,
1730 uint segstart, uint segend, int exidx, char phase, bool gff3) {
1731 static char scorestr[14];
1732 strcpy(scorestr,".");
1733 GffAttrs* xattrs=NULL;
1734 if (exidx>=0) {
1735 if (exons[exidx]->score) sprintf(scorestr,"%.2f", exons[exidx]->score);
1736 xattrs=exons[exidx]->attrs;
1737 }
1738 if (phase==0 || !iscds) phase='.';
1739 const char* ftype=iscds ? "CDS" : getSubfName();
1740 if (gff3) {
1741 fprintf(fout,
1742 "%s\t%s\t%s\t%d\t%d\t%s\t%c\t%c\tParent=%s",
1743 gseqname, tlabel, ftype, segstart, segend, scorestr, strand,
1744 phase, gffID);
1745 if (xattrs!=NULL) {
1746 for (int i=0;i<xattrs->Count();i++)
1747 fprintf(fout, ";%s=%s",names->attrs.getName(xattrs->Get(i)->attr_id),
1748 xattrs->Get(i)->attr_val);
1749 }
1750 fprintf(fout, "\n");
1751 } //GFF
1752 else {//for GTF -- we print only transcripts
1753 //if (isValidTranscript())
1754 fprintf(fout, "%s\t%s\t%s\t%d\t%d\t%s\t%c\t%c\ttranscript_id \"%s\";",
1755 gseqname, tlabel, ftype, segstart, segend, scorestr, strand, phase, gffID);
1756 //char* geneid=(geneID!=NULL)? geneID : gffID;
1757 if (geneID)
1758 fprintf(fout," gene_id \"%s\";",geneID);
1759 if (gene_name!=NULL) {
1760 //fprintf(fout, " gene_name ");
1761 //if (gene_name[0]=='"') fprintf (fout, "%s;",gene_name);
1762 // else fprintf(fout, "\"%s\";",gene_name);
1763 fprintf(fout," gene_name \"%s\";",gene_name);
1764 }
1765 if (xattrs!=NULL) {
1766 for (int i=0;i<xattrs->Count();i++) {
1767 if (xattrs->Get(i)->attr_val==NULL) continue;
1768 const char* attrname=names->attrs.getName(xattrs->Get(i)->attr_id);
1769 fprintf(fout, " %s ",attrname);
1770 if (xattrs->Get(i)->attr_val[0]=='"')
1771 fprintf(fout, "%s;",xattrs->Get(i)->attr_val);
1772 else fprintf(fout, "\"%s\";",xattrs->Get(i)->attr_val);
1773 }
1774 }
1775 //for GTF, also append the GffObj attributes to each exon line
1776 if ((xattrs=this->attrs)!=NULL) {
1777 for (int i=0;i<xattrs->Count();i++) {
1778 if (xattrs->Get(i)->attr_val==NULL) continue;
1779 const char* attrname=names->attrs.getName(xattrs->Get(i)->attr_id);
1780 fprintf(fout, " %s ",attrname);
1781 if (xattrs->Get(i)->attr_val[0]=='"')
1782 fprintf(fout, "%s;",xattrs->Get(i)->attr_val);
1783 else fprintf(fout, "\"%s\";",xattrs->Get(i)->attr_val);
1784 }
1785 }
1786 fprintf(fout, "\n");
1787 }//GTF
1788 }
1789
1790 void GffObj::printGxf(FILE* fout, GffPrintMode gffp,
1791 const char* tlabel, const char* gfparent) {
1792 static char tmpstr[255];
1793 if (tlabel==NULL) {
1794 tlabel=track_id>=0 ? names->tracks.Get(track_id)->name :
1795 (char*)"gffobj" ;
1796 }
1797 unxcoord();
1798 //if (exons.Count()==0) return;
1799 const char* gseqname=names->gseqs.Get(gseq_id)->name;
1800 bool gff3 = (gffp>=pgffAny);
1801 bool showCDS = (gffp==pgtfAny || gffp==pgtfCDS || gffp==pgffCDS || gffp==pgffAny || gffp==pgffBoth);
1802 bool showExon = (gffp<=pgtfExon || gffp==pgffAny || gffp==pgffExon || gffp==pgffBoth);
1803 if (gff3) {
1804 //print GFF3 mRNA line:
1805 if (gscore>0.0) sprintf(tmpstr,"%.2f", gscore);
1806 else strcpy(tmpstr,".");
1807 uint pstart, pend;
1808 if (gffp==pgffCDS) {
1809 pstart=CDstart;
1810 pend=CDend;
1811 }
1812 else { pstart=start;pend=end; }
1813 //const char* ftype=isTranscript() ? "mRNA" : getFeatureName();
1814 const char* ftype=getFeatureName();
1815 fprintf(fout,
1816 "%s\t%s\t%s\t%d\t%d\t%s\t%c\t.\tID=%s",
1817 gseqname, tlabel, ftype, pstart, pend, tmpstr, strand, gffID);
1818 if (CDstart>0 && !showCDS && !isCDS) fprintf(fout,";CDS=%d-%d",CDstart,CDend);
1819 if (gfparent!=NULL) {
1820 //parent override
1821 fprintf(fout, ";Parent=%s",gfparent);
1822 }
1823 else {
1824 if (parent!=NULL && !parent->isDiscarded())
1825 fprintf(fout, ";Parent=%s",parent->getID());
1826 }
1827 if (geneID!=NULL)
1828 fprintf(fout, ";geneID=%s",geneID);
1829 if (gene_name!=NULL)
1830 fprintf(fout, ";gene_name=%s",gene_name);
1831 if (attrs!=NULL) {
1832 for (int i=0;i<attrs->Count();i++) {
1833 const char* attrname=names->attrs.getName(attrs->Get(i)->attr_id);
1834 fprintf(fout,";%s=%s", attrname,
1835 attrs->Get(i)->attr_val);
1836 }
1837 }
1838 fprintf(fout,"\n");
1839 }// gff3 mRNA line
1840 if (showExon) {
1841 //print exons
1842 if (isCDS && exons.Count()>0 &&
1843 ((strand=='-' && exons.Last()->phase<'0') || (strand=='+' && exons.Last()->phase<'0')))
1844 updateExonPhase();
1845
1846 for (int i=0;i<exons.Count();i++) {
1847 printGxfLine(fout, tlabel, gseqname, isCDS, exons[i]->start, exons[i]->end, i, exons[i]->phase, gff3);
1848 }
1849 }//printing exons
1850 if (showCDS && !isCDS && CDstart>0) {
1851 GArray<GffCDSeg> cds(true,true);
1852 getCDSegs(cds);
1853 for (int i=0;i<cds.Count();i++) {
1854 printGxfLine(fout, tlabel, gseqname, true, cds[i].start, cds[i].end, -1, cds[i].phase, gff3);
1855 }
1856 } //showCDS
1857 }
1858
1859 void GffObj::updateExonPhase() {
1860 if (!isCDS) return;
1861 int cdsacc=0;
1862 if (CDphase=='1' || CDphase=='2') {
1863 cdsacc+= 3-(CDphase-'0');
1864 }
1865 if (strand=='-') { //reverse strand
1866 for (int i=exons.Count()-1;i>=0;i--) {
1867 exons[i]->phase='0'+ (3-cdsacc%3)%3;
1868 cdsacc+=exons[i]->end-exons[i]->start+1;
1869 }
1870 }
1871 else { //forward strand
1872 for (int i=0;i<exons.Count();i++) {
1873 exons[i]->phase='0'+ (3-cdsacc%3)%3;
1874 cdsacc+=exons[i]->end-exons[i]->start+1;
1875 }
1876 }
1877 }
1878
1879
1880 void GffObj::getCDSegs(GArray<GffCDSeg>& cds) {
1881 GffCDSeg cdseg;
1882 int cdsacc=0;
1883 if (CDphase=='1' || CDphase=='2') {
1884 cdsacc+= 3-(CDphase-'0');
1885 }
1886 if (strand=='-') {
1887 for (int x=exons.Count()-1;x>=0;x--) {
1888 uint sgstart=exons[x]->start;
1889 uint sgend=exons[x]->end;
1890 if (CDend<sgstart || CDstart>sgend) continue;
1891 if (CDstart>=sgstart && CDstart<=sgend)
1892 sgstart=CDstart; //cdstart within this segment
1893 if (CDend>=sgstart && CDend<=sgend)
1894 sgend=CDend; //cdend within this segment
1895 cdseg.start=sgstart;
1896 cdseg.end=sgend;
1897 cdseg.exonidx=x;
1898 //cdseg.phase='0'+(cdsacc>0 ? (3-cdsacc%3)%3 : 0);
1899 cdseg.phase='0'+ (3-cdsacc%3)%3;
1900 cdsacc+=sgend-sgstart+1;
1901 cds.Add(cdseg);
1902 } //for each exon
1903 } // - strand
1904 else { // + strand
1905 for (int x=0;x<exons.Count();x++) {
1906 uint sgstart=exons[x]->start;
1907 uint sgend=exons[x]->end;
1908 if (CDend<sgstart || CDstart>sgend) continue;
1909 if (CDstart>=sgstart && CDstart<=sgend)
1910 sgstart=CDstart; //seqstart within this segment
1911 if (CDend>=sgstart && CDend<=sgend)
1912 sgend=CDend; //seqend within this segment
1913 cdseg.start=sgstart;
1914 cdseg.end=sgend;
1915 cdseg.exonidx=x;
1916 //cdseg.phase='0'+(cdsacc>0 ? (3-cdsacc%3)%3 : 0);
1917 cdseg.phase='0' + (3-cdsacc%3)%3 ;
1918 cdsacc+=sgend-sgstart+1;
1919 cds.Add(cdseg);
1920 } //for each exon
1921 } // + strand
1922 }