ViewVC Help
View File | Revision Log | Show Annotations | View Changeset | Root Listing
root/gclib/gclib/gff.cpp
Revision: 50
Committed: Thu Sep 8 00:47:22 2011 UTC (7 years, 10 months ago) by gpertea
File size: 64571 byte(s)
Log Message:
fixed a loading bug when a CDS was given to a monoFeature() with a self-exon

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;
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_FROM_GFF3 = 0x00000010;
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* pre, 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 lpre=strlen(pre);
60 char cend=pre[lpre-1];
61 char* pos = (caseStrict) ? strstr(info, pre) : strifind(info, pre);
62 if (pos==NULL) return NULL;
63 char* findstart=info;
64 //require word boundary on the left:
65 while (pos!=NULL && pos!=info && *(pos-1)!=';' && *(pos-1)!=' ') {
66 findstart=pos+lpre;
67 pos = (caseStrict) ? strstr(findstart, pre) : strifind(findstart, pre);
68 }
69 if (pos==NULL) return NULL;
70 if (cend!=' ' && cend!='=') {
71 //require word boundary on the right:
72 while (pos!=NULL && *(pos+lpre)!=' ' && *(pos+lpre)!='=') {
73 findstart=pos+lpre;
74 pos = (caseStrict) ? strstr(findstart, pre) : strifind(findstart, pre);
75 }
76 }
77 if (pos==NULL) return NULL;
78 char* vp=pos+lpre;
79 while (*vp==' ') vp++;
80 if (*vp==';' || *vp==0)
81 GError("Error parsing value of GFF attribute \"%s\", line:\n%s\n", pre, dupline);
82 bool dq_enclosed=false; //value string enclosed by double quotes
83 if (*vp=='"') {
84 dq_enclosed=true;
85 vp++;
86 }
87 if (enforce_GTF2 && !dq_enclosed)
88 GError(GTF2_ERR,pre, dupline);
89 char* vend=vp;
90 if (dq_enclosed) {
91 while (*vend!='"' && *vend!=';' && *vend!=0) vend++;
92 }
93 else {
94 while (*vend!=';' && *vend!=0) vend++;
95 }
96 if (enforce_GTF2 && *vend!='"')
97 GError(GTF2_ERR, pre, dupline);
98 char *r=Gstrdup(vp, vend-1);
99 //-- now remove this attribute from the info string
100 while (*vend!=0 && (*vend=='"' || *vend==';' || *vend==' ')) vend++;
101 if (*vend==0) vend--;
102 for (char *src=vend, *dest=pos;;src++,dest++) {
103 *dest=*src;
104 if (*src==0) break;
105 }
106 return r;
107 }
108
109 static char fnamelc[128];
110
111 GffLine::GffLine(GffReader* reader, const char* l) {
112 llen=strlen(l);
113 GMALLOC(line,llen+1);
114 memcpy(line, l, llen+1);
115 GMALLOC(dupline, llen+1);
116 memcpy(dupline, l, llen+1);
117 skip=true;
118 gseqname=NULL;
119 track=NULL;
120 ftype=NULL;
121 info=NULL;
122 _parents=NULL;
123 _parents_len=0;
124 num_parents=0;
125 parents=NULL;
126 is_gff3=false;
127 is_cds=false;
128 is_transcript=false;
129 is_exon=false;
130 is_gene=false;
131 exontype=0;
132 gene_id=NULL;
133 gene_name=NULL;
134 qstart=0;
135 qend=0;
136 qlen=0;
137 ID=NULL;
138 char* t[9];
139 int i=0;
140 int tidx=1;
141 t[0]=line;
142
143 while (line[i]!=0) {
144 if (line[i]=='\t') {
145 line[i]=0;
146 t[tidx]=line+i+1;
147 tidx++;
148 if (tidx>8) break;
149 }
150 i++;
151 }
152
153 if (tidx<8) { // ignore non-GFF lines
154 // GMessage("Warning: error parsing GFF/GTF line:\n%s\n", l);
155 return;
156 }
157 gseqname=t[0];
158 track=t[1];
159 ftype=t[2];
160 info=t[8];
161 char* p=t[3];
162 if (!parseUInt(p,fstart))
163 GError("Error parsing start coordinate from GFF line:\n%s\n",l);
164 p=t[4];
165 if (!parseUInt(p,fend))
166 GError("Error parsing end coordinate from GFF line:\n%s\n",l);
167 if (fend<fstart) swap(fend,fstart); //make sure fstart>=fend, always
168 p=t[5];
169 if (p[0]=='.' && p[1]==0) {
170 score=0;
171 }
172 else {
173 if (!parseDouble(p,score))
174 GError("Error parsing feature score from GFF line:\n%s\n",l);
175 }
176 strand=*t[6];
177 if (strand!='+' && strand!='-' && strand!='.')
178 GError("Error parsing strand (%c) from GFF line:\n%s\n",strand,l);
179 phase=*t[7]; // must be '.', '0', '1' or '2'
180 ID=NULL;
181 // exon/CDS/mrna filter
182 strncpy(fnamelc, ftype, 127);
183 fnamelc[127]=0;
184 strlower(fnamelc); //convert to lower case
185 bool is_t_data=false;
186 if (strstr(fnamelc, "utr")!=NULL) {
187 exontype=exgffUTR;
188 is_exon=true;
189 is_t_data=true;
190 }
191 else if (strstr(fnamelc, "exon")!=NULL) {
192 exontype=exgffExon;
193 is_exon=true;
194 is_t_data=true;
195 }
196 else if (strstr(fnamelc, "stop") &&
197 (strstr(fnamelc, "codon") || strstr(fnamelc, "cds"))){
198 exontype=exgffStop;
199 is_cds=true; //though some place it outside the last CDS segment
200 is_t_data=true;
201 }
202 else if (strstr(fnamelc, "start") &&
203 ((strstr(fnamelc, "codon")!=NULL) || strstr(fnamelc, "cds")!=NULL)){
204 exontype=exgffStart;
205 is_cds=true;
206 is_t_data=true;
207 }
208 else if (strcmp(fnamelc, "cds")==0) {
209 exontype=exgffCDS;
210 is_cds=true;
211 is_t_data=true;
212 }
213 else if (endsWith(fnamelc, "gene") || startsWith(fnamelc, "gene")) {
214 is_gene=true;
215 is_t_data=true; //because its name will be attached to parented transcripts
216 }
217 else if (endsWith(fnamelc,"rna") || endsWith(fnamelc,"transcript")) {
218 is_transcript=true;
219 is_t_data=true;
220 }
221
222 if (reader->transcriptsOnly && !is_t_data) {
223 char* id=extractAttr("ID=");
224 if (id==NULL) id=extractAttr("transcript_id");
225 //GMessage("Discarding non-transcript line:\n%s\n",l);
226 if (id!=NULL) {
227 reader->discarded_ids.Add(id, new int(1));
228 GFREE(id);
229 }
230 return; //skip this line, unwanted feature name
231 }
232 ID=extractAttr("ID=");
233 char* Parent=extractAttr("Parent=");
234 is_gff3=(ID!=NULL || Parent!=NULL);
235 if (is_gff3) {
236 //parse as GFF3
237 if (ID!=NULL) {
238 //has ID attr so it's likely to be a parent feature
239 //look for explicit gene name
240 gene_name=extractAttr("gene_name=",false);
241 if (gene_name==NULL) {
242 gene_name=extractAttr("geneName=",false);
243 if (gene_name==NULL) {
244 gene_name=extractAttr("gene_sym=",false);
245 if (gene_name==NULL) {
246 gene_name=extractAttr("gene=",false);
247 }
248 }
249 }
250 gene_id=extractAttr("geneID=",false);
251 if (gene_id==NULL) {
252 gene_id=extractAttr("gene_id=",false);
253 }
254 if (is_gene) {
255 //special case: keep the Name and ID attributes of the gene feature
256 if (gene_name==NULL)
257 gene_name=extractAttr("Name=");
258 if (gene_id==NULL) //the ID is also gene_id in this case
259 gene_id=Gstrdup(ID);
260 //skip=false;
261 //return;
262 GFREE(Parent); //TMI, we really don't care about gene Parents?
263 } //gene feature
264 }// has GFF3 ID
265 if (Parent!=NULL) {
266 //keep Parent attr
267 //parse multiple parents
268 num_parents=1;
269 p=Parent;
270 int last_delim_pos=-1;
271 while (*p!=';' && *p!=0) {
272 if (*p==',' && *(p+1)!=0 && *(p+1)!=';') {
273 num_parents++;
274 last_delim_pos=(p-Parent);
275 }
276 p++;
277 }
278 _parents_len=p-Parent+1;
279 _parents=Parent;
280 GMALLOC(parents, num_parents*sizeof(char*));
281 parents[0]=_parents;
282 int i=1;
283 if (last_delim_pos>0) {
284 for (p=_parents+1;p<=_parents+last_delim_pos;p++) {
285 if (*p==',') {
286 char* ep=p-1;
287 while (*ep==' ' && ep>_parents) ep--;
288 *(ep+1)=0; //end the string there
289 parents[i]=p+1;
290 i++;
291 }
292 }
293 }
294 } //has Parent field
295 } //GFF3
296 else { // GTF-like expected
297 Parent=extractAttr("transcript_id");
298 if (Parent!=NULL) { //GTF2 format detected
299 if (is_transcript) {
300 // atypical GTF with a parent transcript line declared
301 ID=Parent;
302 Parent=NULL;
303 }
304 gene_id=extractAttr("gene_id"); // for GTF this is the only attribute accepted as geneID
305 gene_name=extractAttr("gene_name");
306 if (gene_name==NULL) {
307 gene_name=extractAttr("gene_sym");
308 if (gene_name==NULL)
309 gene_name=extractAttr("gene");
310 }
311 //prepare for parseAttr by adding '=' character instead of spaces for all attributes
312 //after the attribute name
313 p=info;
314 bool noed=true; //not edited after the last delim
315 bool nsp=false; //non-space found after last delim
316 while (*p!=0) {
317 if (*p==' ') {
318 if (nsp && noed) {
319 *p='=';
320 noed=false;
321 p++;
322 continue;
323 }
324 }
325 else nsp=true; //non-space
326 if (*p==';') { noed=true; nsp=false; }
327 p++;
328 }
329 } //GTF2 detected (no parent line)
330 else {// Parent is NULL, check for jigsaw format or other pre-GTF2 format
331 //char* fexon=strstr(fnamelc, "exon");
332 //if (fexon!=NULL) {
333 if (exontype==exgffExon) {
334 if (startsWith(track,"jigsaw")) {
335 is_cds=true;
336 strcpy(track,"jigsaw");
337 p=strchr(info,';');
338 if (p==NULL) { Parent=Gstrdup(info); info=NULL; }
339 else { Parent=Gstrdup(info,p-1);
340 info=p+1;
341 }
342 }
343 } //exon feature?
344 if (Parent==NULL && exontype>=exgffCDS &&
345 (i=strcspn(info,"; \t\n\r"))<=(int)(strlen(info)+1)) {
346 //one word ID ? really desperate attempt to parse it here
347 Parent=Gstrdup(info,info+i-1);
348 info=NULL; //discard anything else on the line
349 }
350 }
351 if (Parent!=NULL) { //GTF transcript_id for exon/CDS feature
352 _parents=Parent;
353 GMALLOC(parents,sizeof(char*));
354 num_parents=1;
355 parents[0]=_parents;
356 }
357 } //GTF-like
358
359 //parse other potentially useful features
360 if (is_gff3) {
361 if ((p=strstr(info,"Target="))!=NULL) { //has Target attr
362 p+=7;
363 while (*p!=';' && *p!=0 && *p!=' ') p++;
364 if (*p!=' ') {
365 GError("Error parsing target coordinates from GFF line:\n%s\n",l);
366 }
367 if (!parseUInt(p,qstart))
368 GError("Error parsing target start coordinate from GFF line:\n%s\n",l);
369 if (*p!=' ') {
370 GError("Error parsing next target coordinate from GFF line:\n%s\n",l);
371 }
372 p++;
373 if (!parseUInt(p,qend))
374 GError("Error parsing target end coordinate from GFF line:\n%s\n",l);
375 }
376 if ((p=strifind(info,"Qreg="))!=NULL) { //has Qreg attr
377 p+=5;
378 if (!parseUInt(p,qstart))
379 GError("Error parsing target start coordinate from GFF line:\n%s\n",l);
380 if (*p!='-') {
381 GError("Error parsing next target coordinate from GFF line:\n%s\n",l);
382 }
383 p++;
384 if (!parseUInt(p,qend))
385 GError("Error parsing target end coordinate from GFF line:\n%s\n",l);
386 if (*p=='|' || *p==':') {
387 p++;
388 if (!parseUInt(p,qlen))
389 GError("Error parsing target length from GFF Qreg|: \n%s\n",l);
390 }
391 }//has Qreg attr
392 if (qlen==0 && (p=strifind(info,"Qlen="))!=NULL) {
393 p+=5;
394 if (!parseUInt(p,qlen))
395 GError("Error parsing target length from GFF Qlen:\n%s\n",l);
396 }
397 }//parsing some useful attributes in GFF3 records
398 if (ID==NULL && parents==NULL) {
399 if (reader->gff_warns)
400 GMessage("Warning: could not parse ID or Parent from GFF line:\n%s\n",dupline);
401 return; //skip
402 }
403 skip=false;
404 }
405
406 int GffObj::addExon(GffReader* reader, GffLine* gl, bool keepAttr, bool noExonAttr) {
407 //this will make sure we have the right subftype_id!
408 int subf_id=-1;
409 if (!isTranscript() && gl->is_cds && monoFeature()) {
410 isTranscript(true);
411 exon_ftype_id=gff_fid_exon;
412 if (exons.Count()==1) exons[0]->exontype=exgffExon;
413 }
414 if (isTranscript()) {
415 if (exon_ftype_id<0) {//exon_ftype_id=gff_fid_exon;
416 if (gl->exontype>0) exon_ftype_id=gff_fid_exon;
417 else exon_ftype_id=names->feats.addName(gl->ftype);
418 }
419 //any recognized mRNA segment gets the generic "exon" type (also applies to CDS)
420 if (gl->exontype==0 && !gl->is_transcript) {
421 //extraneous mRNA feature, discard
422 if (reader->gff_warns)
423 GMessage("Warning: discarding unrecognized transcript subfeature %s of %s\n",
424 gl->ftype, gffID);
425 return -1;
426 }
427 }
428 else { //non-mRNA parent feature, check this subf type
429 subf_id=names->feats.addName(gl->ftype);
430 if (exon_ftype_id<0 || exons.Count()==0) //never assigned a subfeature type before (e.g. first exon being added)
431 exon_ftype_id=subf_id;
432 else {
433 if (exon_ftype_id!=subf_id) {
434 //
435 if (exon_ftype_id==ftype_id && exons.Count()==1 && exons[0]->start==start && exons[0]->end==end) {
436 //the existing exon was just a dummy one created by default, discard it
437 exons.Clear();
438 covlen=0;
439 exon_ftype_id=subf_id; //allow the new subfeature to completely takeover
440 }
441 else { //multiple subfeatures, prefer those with
442 if (reader->gff_warns)
443 GMessage("GFF Warning: multiple subfeatures (%s and %s) found for %s, discarding ",
444 names->feats.getName(subf_id), names->feats.getName(exon_ftype_id),gffID);
445 if (gl->exontype!=0) { //new feature is an exon, discard previously parsed subfeatures
446 if (reader->gff_warns) GMessage("%s.\n", names->feats.getName(exon_ftype_id));
447 exon_ftype_id=subf_id;
448 exons.Clear();
449 covlen=0;
450 }
451 else { //discard new feature
452 if (reader->gff_warns) GMessage("%s.\n", names->feats.getName(subf_id));
453 return -1; //skip this 2nd subfeature type for this parent!
454 }
455 }
456 } //incoming subfeature is of different type
457 } //new subfeature type
458 } //non-mRNA parent
459 int eidx=addExon(gl->fstart, gl->fend, gl->score, gl->phase,
460 gl->qstart,gl->qend, gl->is_cds, gl->exontype);
461 if (eidx<0) return eidx; //this should never happen
462 if (keepAttr) {
463 if (noExonAttr) {
464 if (attrs==NULL) //place the parsed attributes directly at transcript level
465 parseAttrs(attrs, gl->info);
466 }
467 else { //need all exon-level attributes
468 parseAttrs(exons[eidx]->attrs, gl->info, true);
469 }
470 }
471 return eidx;
472 }
473
474
475 int GffObj::addExon(uint segstart, uint segend, double sc, char fr, int qs, int qe, bool iscds, char exontype) {
476 if (exons.Count()==0) {
477 if (iscds) isCDS=true; //for now, assume CDS only if first "exon" given is a CDS
478 if (exon_ftype_id<0) {
479 exon_ftype_id = isTranscript() ? gff_fid_exon : ftype_id;
480 }
481 }
482 //special treatment of start/stop codon features, they might be broken/split between exons
483 //and in that case some providers will still give the wrong end coordinate as start+2 (e.g. UCSC)
484 //so we should not trust the end coordinate for such features
485 if (exontype==exgffStart || exontype==exgffStop) {
486 if (strand=='-') segstart=segend;
487 else segend=segstart;
488 if (exontype==exgffStart) {
489 if (CDstart==0 || segstart<CDstart) CDstart=segstart;
490 }
491 else {
492 if (segstart>CDend) CDend=segstart;
493 }
494 }
495 else if (iscds) { //update CDS anchors:
496 if (CDstart==0 || segstart<CDstart) {
497 CDstart=segstart;
498 if (exontype==exgffCDS && strand=='+') CDphase=fr;
499 }
500 if (segend>CDend) {
501 if (exontype==exgffCDS && strand=='-') CDphase=fr;
502 CDend=segend;
503 }
504 }
505 else { // not a CDS/start/stop
506 isCDS=false;
507 }
508 if (qs || qe) {
509 if (qs>qe) swap(qs,qe);
510 if (qs==0) qs=1;
511 }
512 int ovlen=0;
513 if (exontype>0) { //check for overlaps between exon-type segments
514 int oi=exonOverlapIdx(segstart, segend, &ovlen);
515 if (oi>=0) { //overlap existing segment
516 if (ovlen==0) {
517 //adjacent segments will be merged
518 //e.g. CDS to (UTR|exon)
519 if ((exons[oi]->exontype>=exgffUTR && exontype==exgffCDS) ||
520 (exons[oi]->exontype==exgffCDS && exontype>=exgffUTR)) {
521 expandExon(oi, segstart, segend, exgffCDSUTR, sc, fr, qs, qe);
522 return oi;
523 }
524 //CDS adjacent to stop_codon: UCSC does (did?) this
525 if ((exons[oi]->exontype==exgffStop && exontype==exgffCDS) ||
526 (exons[oi]->exontype==exgffCDS && exontype==exgffStop)) {
527 expandExon(oi, segstart, segend, exgffCDS, sc, fr, qs, qe);
528 return oi;
529 }
530 }
531 //only allow this for CDS within exon, stop_codon within (CDS|UTR|exon),
532 // start_codon within (CDS|exon)
533 if (exons[oi]->exontype>exontype &&
534 exons[oi]->start<=segstart && exons[oi]->end>=segend &&
535 !(exons[oi]->exontype==exgffUTR && exontype==exgffCDS)) {
536 //larger segment given first, now the smaller included one is redundant
537 return oi; //only used to store attributes from current GffLine
538 }
539 if (exontype>exons[oi]->exontype &&
540 segstart<=exons[oi]->start && segend>=exons[oi]->end &&
541 !(exontype==exgffUTR && exons[oi]->exontype==exgffCDS)) {
542 //smaller segment given first, so we have to enlarge it
543 expandExon(oi, segstart, segend, exontype, sc, fr, qs, qe);
544 //this should also check for overlapping next exon (oi+1) ?
545 return oi;
546 }
547 //there is also the special case of "ribosomal slippage exception" (programmed frameshift)
548 //where two CDS segments may actually overlap for 1 or 2 bases, but there should be only one encompassing exon
549 //if (ovlen>2 || exons[oi]->exontype!=exgffCDS || exontype!=exgffCDS) {
550 // had to relax this because of some weird UCSC annotations with exons partially overlapping the CDS segments
551 /*
552 if (ovlen>2 && exons[oi]->exontype!=exgffUTR && exontype!=exgffUTR) {
553 if (gff_show_warnings)
554 GMessage("GFF Warning: discarding overlapping feature segment (%d-%d) (vs %d-%d (%s)) for GFF ID %s on %s\n",
555 segstart, segend, exons[oi]->start, exons[oi]->end, getSubfName(), gffID, getGSeqName());
556 hasErrors(true);
557 return -1; //segment NOT added
558 }
559 */
560
561 if ((ovlen>2 || ovlen==0) || exons[oi]->exontype!=exgffCDS || exontype!=exgffCDS) {
562 if (gff_show_warnings)
563 GMessage("GFF Warning: merging overlapping/adjacent feature segment (%d-%d) into (%d-%d) (%s) for GFF ID %s on %s\n",
564 segstart, segend, exons[oi]->start, exons[oi]->end, getSubfName(), gffID, getGSeqName());
565 expandExon(oi, segstart, segend, exontype, sc, fr, qs, qe);
566 return oi;
567 }
568 // else add the segment if the overlap is small and between two CDS segments
569 //TODO: we might want to add an attribute here with the slippage coordinate and size?
570 covlen-=ovlen;
571 }//overlap or adjacent to existing segment
572 } //check for overlap
573 // --- no overlap, or accepted micro-overlap (ribosomal slippage)
574 // create & add the new segment
575 GffExon* enew=new GffExon(segstart, segend, sc, fr, qs, qe, exontype);
576 int eidx=exons.Add(enew);
577 if (eidx<0) {
578 //this would actually be acceptable if the object is a "Gene" and "exons" are in fact isoforms
579 if (gff_show_warnings)
580 GMessage("GFF Warning: failed adding segment %d-%d for %s (discarded)!\n",
581 segstart, segend, gffID);
582 delete enew;
583 hasErrors(true);
584 return -1;
585 }
586 covlen+=(int)(exons[eidx]->end-exons[eidx]->start)+1;
587 start=exons.First()->start;
588 end=exons.Last()->end;
589 if (uptr!=NULL) { //collect stats about the underlying genomic sequence
590 GSeqStat* gsd=(GSeqStat*)uptr;
591 if (start<gsd->mincoord) gsd->mincoord=start;
592 if (end>gsd->maxcoord) gsd->maxcoord=end;
593 if (this->len()>gsd->maxfeat_len) {
594 gsd->maxfeat_len=this->len();
595 gsd->maxfeat=this;
596 }
597 }
598 return eidx;
599 }
600
601 void GffObj::expandExon(int oi, uint segstart, uint segend, char exontype, double sc, char fr, int qs, int qe) {
602 //oi is the index of the *first* overlapping segment found that must be enlarged
603 covlen-=exons[oi]->len();
604 if (segstart<exons[oi]->start)
605 exons[oi]->start=segstart;
606 if (qs && qs<exons[oi]->qstart) exons[oi]->qstart=qs;
607 if (segend>exons[oi]->end)
608 exons[oi]->end=segend;
609 if (qe && qe>exons[oi]->qend) exons[oi]->qend=qe;
610 //warning: score cannot be properly adjusted! e.g. if it's a p-value it's just going to get worse
611 if (sc!=0) exons[oi]->score=sc;
612 covlen+=exons[oi]->len();
613 //if (exons[oi]->exontype< exontype) -- always true
614 exons[oi]->exontype = exontype;
615 if (exontype==exgffCDS) exons[oi]->phase=fr;
616 //we must check if any more exons are also overlapping this
617 int ni=oi+1; //next exon index after oi
618 while (ni<exons.Count() && segend>=exons[ni]->start) { // next segment overlaps new enlarged segment
619 //only allow this if next segment is fully included, and a subordinate
620 if (exons[ni]->exontype<exontype && exons[ni]->end<=segend) {
621 /* I guess we have to relax this due to stupid UCSC hg18 files having a start_codon sticking out
622 chr1 hg18_knownGene start_codon 69806911 69806913 0.000000 + .
623 chr1 hg18_knownGene CDS 69806911 69806912 0.000000 + 0
624 chr1 hg18_knownGene exon 69805456 69806912 0.000000 + .
625 */
626 if (exons[ni]->qstart<exons[oi]->qstart) exons[oi]->qstart=exons[ni]->qstart;
627 if (exons[ni]->qend>exons[oi]->qend) exons[oi]->qend=exons[ni]->qend;
628 exons.Delete(ni);
629 }
630 else {
631 if (gff_show_warnings) GMessage("GFF Warning: overlapping existing exon(%d-%d) while expanding to %d-%d for GFF ID %s\n",
632 exons[ni]->start, exons[ni]->end, segstart, segend, gffID);
633 //hasErrors(true);
634 break;
635 }
636 }
637 // -- make sure any other related boundaries are updated:
638 start=exons.First()->start;
639 end=exons.Last()->end;
640 if (uptr!=NULL) { //collect stats about the underlying genomic sequence
641 GSeqStat* gsd=(GSeqStat*)uptr;
642 if (start<gsd->mincoord) gsd->mincoord=start;
643 if (end>gsd->maxcoord) gsd->maxcoord=end;
644 if (this->len()>gsd->maxfeat_len) {
645 gsd->maxfeat_len=this->len();
646 gsd->maxfeat=this;
647 }
648 }
649 }
650
651 void GffObj::removeExon(int idx) {
652 /*
653 if (idx==0 && segs[0].start==gstart)
654 gstart=segs[1].start;
655 if (idx==segcount && segs[segcount].end==gend)
656 gend=segs[segcount-1].end;
657 */
658 if (idx<0 || idx>=exons.Count()) return;
659 int segstart=exons[idx]->start;
660 int segend=exons[idx]->end;
661 exons.Delete(idx);
662 covlen -= (int)(segend-segstart)+1;
663 start=exons.First()->start;
664 end=exons.Last()->end;
665 if (isCDS) { CDstart=start; CDend=end; }
666 }
667
668 void GffObj::removeExon(GffExon* p) {
669 for (int idx=0;idx<exons.Count();idx++) {
670 if (exons[idx]==p) {
671 int segstart=exons[idx]->start;
672 int segend=exons[idx]->end;
673 exons.Delete(idx);
674 covlen -= (int)(segend-segstart)+1;
675 start=exons.First()->start;
676 end=exons.Last()->end;
677 if (isCDS) { CDstart=start; CDend=end; }
678 return;
679 }
680 }
681 }
682
683
684
685 GffObj::GffObj(GffReader *gfrd, GffLine* gffline, bool keepAttr, bool noExonAttr):
686 GSeg(0,0), exons(true,true,false), children(1,false) {
687 xstart=0;
688 xend=0;
689 xstatus=0;
690 partial=false;
691 isCDS=false;
692 uptr=NULL;
693 ulink=NULL;
694 parent=NULL;
695 udata=0;
696 flags=0;
697 CDstart=0;
698 CDend=0;
699 CDphase=0;
700 geneID=NULL;
701 gene_name=NULL;
702 attrs=NULL;
703 gffID=NULL;
704 track_id=-1;
705 gseq_id=-1;
706 ftype_id=-1;
707 exon_ftype_id=-1;
708 strand='.';
709 if (gfrd==NULL)
710 GError("Cannot use this GffObj constructor with a NULL GffReader!\n");
711 gffnames_ref(names);
712 if (gfrd->names==NULL) gfrd->names=names;
713 //qlen=0;qstart=0;qend=0;
714 gscore=0;
715 uscore=0;
716 covlen=0;
717 qcov=0;
718 start=gffline->fstart;
719 end=gffline->fend;
720 gseq_id=names->gseqs.addName(gffline->gseqname);
721 track_id=names->tracks.addName(gffline->track);
722 strand=gffline->strand;
723 qlen=gffline->qlen;
724 qstart=gffline->qstart;
725 qend=gffline->qend;
726 //setup flags from gffline
727 isCDS=gffline->is_cds; //for now
728 isGene(gffline->is_gene);
729 isTranscript(gffline->is_transcript || gffline->exontype!=0);
730 fromGff3(gffline->is_gff3);
731
732 if (gffline->parents!=NULL) {
733 //GTF style -- create a GffObj directly by subfeature
734 //(also possible orphan GFF3 exon line, or an exon given before its parent (chado))
735 if (gffline->exontype!=0) { //recognized exon-like feature
736 ftype_id=gff_fid_transcript; //so this is some sort of transcript
737 exon_ftype_id=gff_fid_exon; //subfeatures MUST be exons
738 }
739 else {//unrecognized subfeatures
740 //make this GffObj of the same feature type
741 ftype_id=names->feats.addName(gffline->ftype);
742 }
743 if (gffline->ID==NULL) { //typical GTF
744 gffID=Gstrdup(gffline->parents[0]);
745 this->createdByExon(true);
746 //this is likely the first exon/segment of the feature
747 addExon(gfrd, gffline, keepAttr, noExonAttr);
748 }
749 else { //a parented feature with an ID -- probably an orphan GFF3 line
750 if (gffline->is_gff3 && gffline->exontype!=0) {
751 //premature exon given before its parent transcript
752 //create the transcript entry here
753 gffID=Gstrdup(gffline->parents[0]);
754 this->createdByExon(true);
755 //this is the first exon/segment of the transcript
756 addExon(gfrd, gffline, keepAttr, noExonAttr);
757 }
758 else { //unrecognized non-exon feature ? use the ID instead
759 gffID=Gstrdup(gffline->ID);
760 if (keepAttr) this->parseAttrs(attrs, gffline->info);
761 }
762 }
763 } //subfeature given directly
764 else { //gffline->parents==NULL
765 //create a parent feature in its own right
766 gscore=gffline->score;
767 if (gffline->ID==NULL || gffline->ID[0]==0)
768 GError("Error: no ID found for GFF record start\n");
769 gffID=Gstrdup(gffline->ID); //there must be an ID here
770 //if (gffline->is_transcript) ftype_id=gff_fid_mRNA;
771 //else
772 ftype_id=names->feats.addName(gffline->ftype);
773 if (gffline->is_transcript)
774 exon_ftype_id=gff_fid_exon;
775
776 if (keepAttr) this->parseAttrs(attrs, gffline->info);
777 }//no parent
778
779 if (gffline->gene_name!=NULL) {
780 gene_name=Gstrdup(gffline->gene_name);
781 }
782 if (gffline->gene_id!=NULL) {
783 geneID=Gstrdup(gffline->gene_id);
784 }
785
786 GSeqStat* gsd=gfrd->gseqstats.AddIfNew(new GSeqStat(gseq_id,names->gseqs.lastNameUsed()),true);
787 uptr=gsd;
788 if (start<gsd->mincoord) gsd->mincoord=start;
789 if (end>gsd->maxcoord) gsd->maxcoord=end;
790 if (this->len()>gsd->maxfeat_len) {
791 gsd->maxfeat_len=this->len();
792 gsd->maxfeat=this;
793 }
794 }
795
796 GffLine* GffReader::nextGffLine() {
797 if (gffline!=NULL) return gffline; //caller should free gffline after processing
798 while (gffline==NULL) {
799 int llen=0;
800 buflen=GFF_LINELEN-1;
801 char* l=fgetline(linebuf, buflen, fh, &fpos, &llen);
802 if (l==NULL) {
803 return NULL; //end of file
804 }
805 int ns=0; //first nonspace position
806 while (l[ns]!=0 && isspace(l[ns])) ns++;
807 if (l[ns]=='#' || llen<10) continue;
808 gffline=new GffLine(this, l);
809 if (gffline->skip) {
810 delete gffline;
811 gffline=NULL;
812 continue;
813 }
814 if (gffline->ID==NULL && gffline->parents==NULL) { //it must have an ID
815 //this might not be needed, already checked in the GffLine constructor
816 if (gff_warns)
817 GMessage("Warning: malformed GFF line, no parent or record Id (kipping\n");
818 delete gffline;
819 gffline=NULL;
820 //continue;
821 }
822 }
823 return gffline;
824 }
825
826 char* GffReader::gfoBuildId(const char* id, const char* ctg) {
827 //caller must free the returned pointer
828 char* buf=NULL;
829 int idlen=strlen(id);
830 GMALLOC(buf, idlen+strlen(ctg)+2);
831 strcpy(buf, id);
832 buf[idlen]='~';
833 strcpy(buf+idlen+1, ctg);
834 return buf;
835 }
836
837 void GffReader::gfoRemove(const char* id, const char* ctg) {
838 char* buf=gfoBuildId(id,ctg);
839 phash.Remove(buf);
840 GFREE(buf);
841 }
842
843 //Warning: if gflst gets altered, idx becomes obsolete
844 GfoHolder* GffReader::gfoAdd(const char* id, const char* ctg, GffObj* gfo, int idx) {
845 char* buf=gfoBuildId(id,ctg);
846 GfoHolder* r=new GfoHolder(gfo,idx);
847 phash.Add(buf, r);
848 GFREE(buf);
849 return r;
850 }
851
852 GfoHolder* GffReader::gfoFind(const char* id, const char* ctg) {
853 char* buf=gfoBuildId(id,ctg);
854 GfoHolder* r=phash.Find(buf);
855 GFREE(buf);
856 return r;
857 }
858
859 GfoHolder* GffReader::replaceGffRec(GffLine* gffline, bool keepAttr, bool noExonAttr, int replaceidx) {
860 GffObj* newgfo=new GffObj(this, gffline, keepAttr, noExonAttr);
861 GfoHolder* r=NULL;
862 if (replaceidx>=0) {
863 gflst.Put(replaceidx,newgfo);
864 r=gfoAdd(newgfo->gffID, gffline->gseqname, newgfo, replaceidx);
865 }
866 else {
867 int gfoidx=gflst.Add(newgfo);
868 r=gfoAdd(newgfo->gffID, gffline->gseqname, newgfo, gfoidx);
869 }
870 if (gff_warns) {
871 int* pcount=tids.Find(newgfo->gffID);
872 if (pcount!=NULL) {
873 if (gff_warns) GMessage("Warning: duplicate GFF ID: %s\n", newgfo->gffID);
874 (*pcount)++;
875 }
876 else {
877 tids.Add(newgfo->gffID,new int(1));
878 }
879 }
880 return r;
881 }
882
883 GfoHolder* GffReader::updateParent(GfoHolder* newgfh, GffObj* parent) {
884 //assert(parent);
885 //assert(newgfo);
886 parent->children.Add(newgfh->gffobj);
887 if (newgfh->gffobj->parent==NULL) newgfh->gffobj->parent=parent;
888 newgfh->gffobj->setLevel(parent->getLevel()+1);
889 if (parent->isGene()) {
890 if (parent->gene_name!=NULL && newgfh->gffobj->gene_name==NULL)
891 newgfh->gffobj->gene_name=Gstrdup(parent->gene_name);
892 if (parent->geneID!=NULL && newgfh->gffobj->geneID==NULL)
893 newgfh->gffobj->geneID=Gstrdup(parent->geneID);
894 }
895
896 return newgfh;
897 }
898
899 GfoHolder* GffReader::newGffRec(GffLine* gffline, bool keepAttr, bool noExonAttr,
900 GffObj* parent, GffExon* pexon) {
901 GffObj* newgfo=new GffObj(this, gffline, keepAttr, noExonAttr);
902 GfoHolder* r=NULL;
903 int gfoidx=gflst.Add(newgfo);
904 r=gfoAdd(newgfo->gffID, gffline->gseqname, newgfo, gfoidx);
905 if (parent!=NULL) {
906 updateParent(r, parent);
907 if (pexon!=NULL) parent->removeExon(pexon);
908 }
909 if (gff_warns) {
910 int* pcount=tids.Find(newgfo->gffID);
911 if (pcount!=NULL) {
912 if (gff_warns) GMessage("Warning: duplicate GFF ID: %s\n", newgfo->gffID);
913 (*pcount)++;
914 }
915 else {
916 tids.Add(newgfo->gffID,new int(1));
917 }
918 }
919 return r;
920 }
921
922 GfoHolder* GffReader::updateGffRec(GfoHolder* prevgfo, GffLine* gffline,
923 bool keepAttr) {
924 if (prevgfo==NULL) return NULL;
925 prevgfo->gffobj->createdByExon(false);
926 prevgfo->gffobj->ftype_id=prevgfo->gffobj->names->feats.addName(gffline->ftype);
927 prevgfo->gffobj->start=gffline->fstart;
928 prevgfo->gffobj->end=gffline->fend;
929 prevgfo->gffobj->isGene(gffline->is_gene);
930 prevgfo->gffobj->isTranscript(gffline->is_transcript || gffline->exontype!=0);
931 prevgfo->gffobj->fromGff3(gffline->is_gff3);
932 if (keepAttr) {
933 if (prevgfo->gffobj->attrs!=NULL) prevgfo->gffobj->attrs->Clear();
934 prevgfo->gffobj->parseAttrs(prevgfo->gffobj->attrs, gffline->info);
935 }
936 return prevgfo;
937 }
938
939
940 bool GffReader::addExonFeature(GfoHolder* prevgfo, GffLine* gffline, GHash<CNonExon>& pex, bool noExonAttr) {
941 bool r=true;
942 if (gffline->strand!=prevgfo->gffobj->strand) {
943 GMessage("GFF Error: duplicate GFF ID '%s' (exons found on different strands of %s)\n",
944 prevgfo->gffobj->gffID, prevgfo->gffobj->getGSeqName());
945 r=false;
946 }
947 int gdist=(gffline->fstart>prevgfo->gffobj->end) ? gffline->fstart-prevgfo->gffobj->end :
948 ((gffline->fend<prevgfo->gffobj->start)? prevgfo->gffobj->start-gffline->fend :
949 0 );
950 if (gdist>(int)GFF_MAX_LOCUS) { //too far apart, most likely this is a duplicate ID
951 GMessage("Error: duplicate GFF ID '%s' (or exons too far apart)!\n",prevgfo->gffobj->gffID);
952 //validation_errors = true;
953 r=false;
954 if (!gff_warns) exit(1);
955 }
956 int eidx=prevgfo->gffobj->addExon(this, gffline, !noExonAttr, noExonAttr);
957 if (eidx>=0 && gffline->ID!=NULL && gffline->exontype==0)
958 subfPoolAdd(pex, prevgfo);
959 return r;
960 }
961
962 CNonExon* GffReader::subfPoolCheck(GffLine* gffline, GHash<CNonExon>& pex, char*& subp_name) {
963 CNonExon* subp=NULL;
964 subp_name=NULL;
965 for (int i=0;i<gffline->num_parents;i++) {
966 if (transcriptsOnly && discarded_ids.Find(gffline->parents[i])!=NULL)
967 continue;
968 subp_name=gfoBuildId(gffline->parents[i], gffline->gseqname); //e.g. mRNA name
969 subp=pex.Find(subp_name);
970 if (subp!=NULL)
971 return subp;
972 GFREE(subp_name);
973 }
974 return NULL;
975 }
976
977 void GffReader::subfPoolAdd(GHash<CNonExon>& pex, GfoHolder* newgfo) {
978 //this might become a parent feature later
979 if (newgfo->gffobj->exons.Count()>0) {
980 char* xbuf=gfoBuildId(gffline->ID, gffline->gseqname);
981 pex.Add(xbuf, new CNonExon(newgfo->idx, newgfo->gffobj,
982 newgfo->gffobj->exons[0], gffline));
983 GFREE(xbuf);
984 }
985 }
986
987 GfoHolder* GffReader::promoteFeature(CNonExon* subp, char*& subp_name, GHash<CNonExon>& pex,
988 bool keepAttr, bool noExonAttr) {
989 GffObj* prevp=subp->parent; //grandparent of gffline (e.g. gene)
990 if (prevp!=gflst[subp->idx])
991 GError("Error promoting subfeature %s, gflst index mismatch?!\n", subp->gffline->ID);
992 subp->gffline->discardParent();
993 GfoHolder* gfoh=newGffRec(subp->gffline, keepAttr, noExonAttr, prevp, subp->exon);
994 pex.Remove(subp_name); //no longer a potential parent, moved it to phash already
995 prevp->promotedChildren(true);
996 return gfoh; //returns the holder of newly promoted feature
997 }
998
999 //have to parse the whole file because exons can be scattered all over
1000 void GffReader::readAll(bool keepAttr, bool mergeCloseExons, bool noExonAttr) {
1001 bool validation_errors = false;
1002 //loc_debug=false;
1003 GHash<CNonExon> pex; //keep track of any "exon"-like features that have an ID
1004 //and thus could become promoted to parent features
1005 while (nextGffLine()!=NULL) {
1006 //seen this gff ID before?
1007 GfoHolder* prevseen=NULL;
1008 if (gffline->ID) //GFF3
1009 prevseen=gfoFind(gffline->ID, gffline->gseqname);
1010 if (prevseen!=NULL) {
1011 if (prevseen->gffobj->createdByExon()) {
1012 updateGffRec(prevseen, gffline, keepAttr);
1013 }
1014 else {
1015 GMessage("Error: duplicate GFF ID '%s' encountered!\n",gffline->ID);
1016 validation_errors = true;
1017 if (gff_warns) {
1018 delete gffline; gffline=NULL; continue;
1019 }
1020 else exit(1);
1021 }
1022 }
1023 if (gffline->parents==NULL) {//start GFF3-like record with no parent (mRNA, gene)
1024 if (!prevseen) newGffRec(gffline, keepAttr, noExonAttr);
1025 }
1026 else { //--- it's a parented feature (could still be a mRNA)
1027 bool found_parent=false;
1028 GfoHolder* newgfo=prevseen;
1029 for (int i=0;i<gffline->num_parents;i++) {
1030 if (transcriptsOnly && discarded_ids.Find(gffline->parents[i])!=NULL)
1031 continue; //skipping discarded parent feature
1032 GfoHolder* parentgfo=gfoFind(gffline->parents[i], gffline->gseqname);
1033 if (parentgfo!=NULL) { //parent GffObj parsed earlier
1034 found_parent=true;
1035 if (parentgfo->gffobj->isGene() && gffline->is_transcript
1036 && gffline->exontype==0) {
1037 //not an exon, but a transcript parented by a gene
1038 if (newgfo) {
1039 updateParent(newgfo, parentgfo->gffobj);
1040 }
1041 else {
1042 newgfo=newGffRec(gffline, keepAttr, noExonAttr, parentgfo->gffobj);
1043 }
1044 }
1045 else { //potential exon subfeature
1046 if (!addExonFeature(parentgfo, gffline, pex, noExonAttr))
1047 validation_errors=true;
1048 }
1049 }
1050 } //for each parsed parent Id
1051 if (!found_parent) { //new GTF-like record starting here with a subfeature directly
1052 //or it could be some chado GFF3 barf with exons declared BEFORE their parent :(
1053 //check if this feature isn't parented by a previously stored "exon" subfeature
1054 char* subp_name=NULL;
1055 CNonExon* subp=subfPoolCheck(gffline, pex, subp_name);
1056 if (subp!=NULL) { //found a subfeature that is the parent of this gffline
1057 //promote that subfeature to a full GffObj
1058 GfoHolder* gfoh=promoteFeature(subp, subp_name, pex, keepAttr, noExonAttr);
1059 //add current gffline as an exon of the newly promoted subfeature
1060 if (!addExonFeature(gfoh, gffline, pex, noExonAttr))
1061 validation_errors=true;
1062 }
1063 else { //no parent seen before, create one directly with this exon
1064 //loc_debug=true;
1065 GfoHolder* newgfo=prevseen ? prevseen : newGffRec(gffline, keepAttr, noExonAttr);
1066 if (gffline->ID!=NULL && gffline->exontype==0)
1067 subfPoolAdd(pex, newgfo);
1068 //even those with errors will be added here!
1069 }
1070 GFREE(subp_name);
1071 } //no previous parent found
1072 } //parented feature
1073 //--
1074 delete gffline;
1075 gffline=NULL;
1076 }//while gff lines
1077 gflst.finalize(this, mergeCloseExons, keepAttr, noExonAttr); //force sorting by locus if so constructed
1078 // all gff records are now loaded in GList gflst
1079 // so we can free the hash
1080 phash.Clear();
1081 tids.Clear();
1082 if (validation_errors) {
1083 exit(1);
1084 }
1085 }
1086
1087 GffObj* GffObj::finalize(GffReader* gfr, bool mergeCloseExons, bool keepAttrs, bool noExonAttr) {
1088 //merge
1089 //always merge adjacent or overlapping segments
1090 //but if mergeCloseExons then merge even when distance is up to 5 bases
1091 udata=0;
1092 uptr=NULL;
1093 if (gfr->transcriptsOnly && !(isTranscript() || (isGene() && children.Count()==0))) {
1094 isDiscarded(true);
1095 }
1096 if (ftype_id==gff_fid_transcript && CDstart>0) {
1097 ftype_id=gff_fid_mRNA;
1098 //exon_ftype_id=gff_fid_exon;
1099 }
1100 //if (ftype_id==gff_fid_mRNA || exon_ftype_id==gff_fid_exon || mergeCloseExons) {
1101 if (isTranscript() || exon_ftype_id==gff_fid_exon || mergeCloseExons) {
1102 int mindist=mergeCloseExons ? 5:1;
1103 for (int i=0;i<exons.Count()-1;i++) {
1104 int ni=i+1;
1105 uint mend=exons[i]->end;
1106 while (ni<exons.Count()) {
1107 int dist=(int)(exons[ni]->start-mend);
1108 if (dist>mindist) break; //no merging with next segment
1109 if (gfr!=NULL && gfr->gff_warns && dist!=0 && (exons[ni]->exontype!=exgffUTR && exons[i]->exontype!=exgffUTR)) {
1110 GMessage("GFF warning: merging adjacent/overlapping segments of %s on %s (%d-%d, %d-%d)\n",
1111 gffID, getGSeqName(), exons[i]->start, exons[i]->end,exons[ni]->start, exons[ni]->end);
1112 }
1113 mend=exons[ni]->end;
1114 covlen-=exons[i]->len();
1115 exons[i]->end=mend;
1116 covlen+=exons[i]->len();
1117 covlen-=exons[ni]->len();
1118 if (exons[ni]->attrs!=NULL && (exons[i]->attrs==NULL ||
1119 exons[i]->attrs->Count()<exons[ni]->attrs->Count())) {
1120 //use the other exon attributes, if more
1121 delete(exons[i]->attrs);
1122 exons[i]->attrs=exons[ni]->attrs;
1123 exons[ni]->attrs=NULL;
1124 }
1125 exons.Delete(ni);
1126 } //check for merge with next exon
1127 } //for each exon
1128 }
1129 //attribute reduction for GTF records
1130 if (keepAttrs && !noExonAttr && !fromGff3()
1131 && exons.Count()>0 && exons[0]->attrs!=NULL) {
1132 bool attrs_discarded=false;
1133 for (int a=0;a<exons[0]->attrs->Count();a++) {
1134 int attr_name_id=exons[0]->attrs->Get(a)->attr_id;
1135 char* attr_name=names->attrs.getName(attr_name_id);
1136 char* attr_val =exons[0]->attrs->Get(a)->attr_val;
1137 bool sameExonAttr=true;
1138 for (int i=1;i<exons.Count();i++) {
1139 char* ov=exons[i]->getAttr(attr_name_id);
1140 if (ov==NULL || (strcmp(ov,attr_val)!=0)) {
1141 sameExonAttr=false;
1142 break;
1143 }
1144 }
1145 if (sameExonAttr) {
1146 //delete this attribute from exons level
1147 attrs_discarded=true;
1148 this->addAttr(attr_name, attr_val);
1149 for (int i=1;i<exons.Count();i++) {
1150 removeExonAttr(*(exons[i]), attr_name_id);
1151 }
1152 exons[0]->attrs->freeItem(a);
1153 }
1154 }
1155 if (attrs_discarded) exons[0]->attrs->Pack();
1156 }
1157 return this;
1158 }
1159
1160 void GffObj::parseAttrs(GffAttrs*& atrlist, char* info, bool isExon) {
1161 if (names==NULL)
1162 GError(ERR_NULL_GFNAMES, "parseAttrs()");
1163 if (atrlist==NULL)
1164 atrlist=new GffAttrs();
1165 char* endinfo=info+strlen(info);
1166 char* start=info;
1167 char* pch=start;
1168 while (start<endinfo) {
1169 //skip spaces
1170 while (*start==' ' && start<endinfo) start++;
1171 pch=strchr(start, ';');
1172 if (pch==NULL) pch=endinfo;
1173 else {
1174 *pch='\0';
1175 pch++;
1176 }
1177 char* ech=strchr(start,'=');
1178 if (ech!=NULL) { // attr=value format found
1179 *ech='\0';
1180 //if (noExonAttr && (strcmp(start, "exon_number")==0 || strcmp(start, "exon")==0)) { start=pch; continue; }
1181 if (strcmp(start, "exon_number")==0 || strcmp(start, "exon")==0 ||
1182 strcmp(start, "exon_id")==0)
1183 { start=pch; continue; }
1184 ech++;
1185 while (*ech==' ' && ech<endinfo) ech++;//skip extra spaces after the '='
1186 //atrlist->Add(new GffAttr(names->attrs.addName(start),ech));
1187 //make sure we don't add the same attribute more than once
1188 if (isExon && (strcmp(start, "protein_id")==0)) {
1189 //Ensembl special case
1190 this->addAttr(start, ech);
1191 start=pch;
1192 continue;
1193 }
1194 atrlist->add_or_update(names, start, ech);
1195 }
1196 /*
1197 else { //not an attr=value format
1198 atrlist->Add(new GffAttr(names->attrs.addName(start),"1"));
1199 }
1200 */
1201 start=pch;
1202 }
1203 if (atrlist->Count()==0) { delete atrlist; atrlist=NULL; }
1204 }
1205
1206 void GffObj::addAttr(const char* attrname, const char* attrvalue) {
1207 if (this->attrs==NULL)
1208 this->attrs=new GffAttrs();
1209 //this->attrs->Add(new GffAttr(names->attrs.addName(attrname),attrvalue));
1210 this->attrs->add_or_update(names, attrname, attrvalue);
1211 }
1212
1213
1214 void GffObj::setFeatureName(const char* feature) {
1215 //change the feature name/type for a transcript
1216 int fid=names->feats.addName(feature);
1217 if (monoFeature() && exons.Count()>0)
1218 this->exon_ftype_id=fid;
1219 this->ftype_id=fid;
1220 }
1221
1222 void GffObj::setRefName(const char* newname) {
1223 //change the feature name/type for a transcript
1224 int rid=names->gseqs.addName(newname);
1225 this->gseq_id=rid;
1226 }
1227
1228
1229
1230 int GffObj::removeAttr(const char* attrname, const char* attrval) {
1231 if (this->attrs==NULL || attrname==NULL || attrname[0]==0) return 0;
1232 int aid=this->names->attrs.getId(attrname);
1233 if (aid<0) return 0;
1234 int delcount=0; //could be more than one ?
1235 for (int i=0;i<this->attrs->Count();i++) {
1236 if (aid==this->attrs->Get(i)->attr_id) {
1237 if (attrval==NULL ||
1238 strcmp(attrval, this->attrs->Get(i)->attr_val)==0) {
1239 delcount++;
1240 this->attrs->freeItem(i);
1241 }
1242 }
1243 }
1244 if (delcount>0) this->attrs->Pack();
1245 return delcount;
1246 }
1247
1248 int GffObj::removeAttr(int aid, const char* attrval) {
1249 if (this->attrs==NULL || aid<0) return 0;
1250 int delcount=0; //could be more than one ?
1251 for (int i=0;i<this->attrs->Count();i++) {
1252 if (aid==this->attrs->Get(i)->attr_id) {
1253 if (attrval==NULL ||
1254 strcmp(attrval, this->attrs->Get(i)->attr_val)==0) {
1255 delcount++;
1256 this->attrs->freeItem(i);
1257 }
1258 }
1259 }
1260 if (delcount>0) this->attrs->Pack();
1261 return delcount;
1262 }
1263
1264
1265 int GffObj::removeExonAttr(GffExon& exon, const char* attrname, const char* attrval) {
1266 if (exon.attrs==NULL || attrname==NULL || attrname[0]==0) return 0;
1267 int aid=this->names->attrs.getId(attrname);
1268 if (aid<0) return 0;
1269 int delcount=0; //could be more than one
1270 for (int i=0;i<exon.attrs->Count();i++) {
1271 if (aid==exon.attrs->Get(i)->attr_id) {
1272 if (attrval==NULL ||
1273 strcmp(attrval, exon.attrs->Get(i)->attr_val)==0) {
1274 delcount++;
1275 exon.attrs->freeItem(i);
1276 }
1277 }
1278 }
1279 if (delcount>0) exon.attrs->Pack();
1280 return delcount;
1281 }
1282
1283 int GffObj::removeExonAttr(GffExon& exon, int aid, const char* attrval) {
1284 if (exon.attrs==NULL || aid<0) return 0;
1285 int delcount=0; //could be more than one
1286 for (int i=0;i<exon.attrs->Count();i++) {
1287 if (aid==exon.attrs->Get(i)->attr_id) {
1288 if (attrval==NULL ||
1289 strcmp(attrval, exon.attrs->Get(i)->attr_val)==0) {
1290 delcount++;
1291 exon.attrs->freeItem(i);
1292 }
1293 }
1294 }
1295 if (delcount>0) exon.attrs->Pack();
1296 return delcount;
1297 }
1298
1299
1300 void GffObj::getCDS_ends(uint& cds_start, uint& cds_end) {
1301 cds_start=0;
1302 cds_end=0;
1303 if (CDstart==0 || CDend==0) return; //no CDS info
1304 int cdsadj=0;
1305 if (CDphase=='1' || CDphase=='2') {
1306 cdsadj=CDphase-'0';
1307 }
1308 cds_start=CDstart;
1309 cds_end=CDend;
1310 if (strand=='-') cds_end-=cdsadj;
1311 else cds_start+=cdsadj;
1312 }
1313
1314 void GffObj::mRNA_CDS_coords(uint& cds_mstart, uint& cds_mend) {
1315 //sets cds_start and cds_end to the CDS start,end coordinates on the spliced mRNA transcript
1316 cds_mstart=0;
1317 cds_mend=0;
1318 if (CDstart==0 || CDend==0) return; //no CDS info
1319 //restore normal coordinates, just in case
1320 unxcoord();
1321 int cdsadj=0;
1322 if (CDphase=='1' || CDphase=='2') {
1323 cdsadj=CDphase-'0';
1324 }
1325 /*
1326 uint seqstart=CDstart;
1327 uint seqend=CDend;
1328 */
1329 uint seqstart=exons.First()->start;
1330 uint seqend=exons.Last()->end;
1331 int s=0; //resulting nucleotide counter
1332 if (strand=='-') {
1333 for (int x=exons.Count()-1;x>=0;x--) {
1334 uint sgstart=exons[x]->start;
1335 uint sgend=exons[x]->end;
1336 if (seqend<sgstart || seqstart>sgend) continue;
1337 if (seqstart>=sgstart && seqstart<=sgend)
1338 sgstart=seqstart; //seqstart within this segment
1339 if (seqend>=sgstart && seqend<=sgend)
1340 sgend=seqend; //seqend within this segment
1341 s+=(int)(sgend-sgstart)+1;
1342 if (CDstart>=sgstart && CDstart<=sgend) {
1343 //CDstart in this segment
1344 //and we are getting the whole transcript
1345 cds_mend=s-(int)(CDstart-sgstart);
1346 }
1347 if (CDend>=sgstart && CDend<=sgend) {
1348 //CDstart in this segment
1349 //and we are getting the whole transcript
1350 cds_mstart=s-(int)(CDend-cdsadj-sgstart);
1351 }
1352 } //for each exon
1353 } // - strand
1354 else { // + strand
1355 for (int x=0;x<exons.Count();x++) {
1356 uint sgstart=exons[x]->start;
1357 uint sgend=exons[x]->end;
1358 if (seqend<sgstart || seqstart>sgend) continue;
1359 if (seqstart>=sgstart && seqstart<=sgend)
1360 sgstart=seqstart; //seqstart within this segment
1361 if (seqend>=sgstart && seqend<=sgend)
1362 sgend=seqend; //seqend within this segment
1363 s+=(int)(sgend-sgstart)+1;
1364 /* for (uint i=sgstart;i<=sgend;i++) {
1365 spliced[s]=gsubseq[i-gstart];
1366 s++;
1367 }//for each nt
1368 */
1369 if (CDstart>=sgstart && CDstart<=sgend) {
1370 //CDstart in this segment
1371 cds_mstart=s-(int)(sgend-CDstart-cdsadj);
1372 }
1373 if (CDend>=sgstart && CDend<=sgend) {
1374 //CDend in this segment
1375 cds_mend=s-(int)(sgend-CDend);
1376 }
1377 } //for each exon
1378 } // + strand
1379 //spliced[s]=0;
1380 //if (rlen!=NULL) *rlen=s;
1381 //return spliced;
1382 }
1383
1384 char* GffObj::getUnspliced(GFaSeqGet* faseq, int* rlen, GList<GSeg>* seglst)
1385 {
1386 if (faseq==NULL) { GMessage("Warning: getUnspliced(NULL,.. ) called!\n");
1387 return NULL;
1388 }
1389 //restore normal coordinates:
1390 unxcoord();
1391 if (exons.Count()==0) return NULL;
1392 int fspan=end-start+1;
1393 const char* gsubseq=faseq->subseq(start, fspan);
1394 if (gsubseq==NULL) {
1395 GError("Error getting subseq for %s (%d..%d)!\n", gffID, start, end);
1396 }
1397 char* unspliced=NULL;
1398
1399 int seqstart=exons.First()->start;
1400 int seqend=exons.Last()->end;
1401
1402 int unsplicedlen = 0;
1403
1404 unsplicedlen += seqend - seqstart + 1;
1405
1406 GMALLOC(unspliced, unsplicedlen+1); //allocate more here
1407 //uint seqstart, seqend;
1408
1409 int s = 0; //resulting nucleotide counter
1410 if (strand=='-')
1411 {
1412 if (seglst!=NULL)
1413 seglst->Add(new GSeg(s+1,s+1+seqend-seqstart));
1414 for (int i=seqend;i>=seqstart;i--)
1415 {
1416 unspliced[s] = ntComplement(gsubseq[i-start]);
1417 s++;
1418 }//for each nt
1419 } // - strand
1420 else
1421 { // + strand
1422 if (seglst!=NULL)
1423 seglst->Add(new GSeg(s+1,s+1+seqend-seqstart));
1424 for (int i=seqstart;i<=seqend;i++)
1425 {
1426 unspliced[s]=gsubseq[i-start];
1427 s++;
1428 }//for each nt
1429 } // + strand
1430 //assert(s <= unsplicedlen);
1431 unspliced[s]=0;
1432 if (rlen!=NULL) *rlen=s;
1433 return unspliced;
1434 }
1435
1436 char* GffObj::getSpliced(GFaSeqGet* faseq, bool CDSonly, int* rlen, uint* cds_start, uint* cds_end,
1437 GList<GSeg>* seglst) {
1438 if (CDSonly && CDstart==0) return NULL;
1439 if (faseq==NULL) { GMessage("Warning: getSpliced(NULL,.. ) called!\n");
1440 return NULL;
1441 }
1442 //restore normal coordinates:
1443 unxcoord();
1444 if (exons.Count()==0) return NULL;
1445 int fspan=end-start+1;
1446 const char* gsubseq=faseq->subseq(start, fspan);
1447 if (gsubseq==NULL) {
1448 GError("Error getting subseq for %s (%d..%d)!\n", gffID, start, end);
1449 }
1450 if (fspan<(int)(end-start+1)) { //special case: stop coordinate was extended past the gseq length, must adjust
1451 int endadj=end-start+1-fspan;
1452 uint prevend=end;
1453 end-=endadj;
1454 if (CDend>end) CDend=end;
1455 if (exons.Last()->end>end) {
1456 exons.Last()->end=end; //this could get us into trouble if exon start is also > end
1457 if (exons.Last()->start>exons.Last()->end) {
1458 GError("GffObj::getSpliced() error: improper genomic coordinate %d on %s for %s\n",
1459 prevend,getGSeqName(), getID());
1460 }
1461 covlen-=endadj;
1462 }
1463 }
1464 char* spliced=NULL;
1465 GMALLOC(spliced, covlen+1); //allocate more here
1466 uint seqstart, seqend;
1467 int cdsadj=0;
1468 if (CDphase=='1' || CDphase=='2') {
1469 cdsadj=CDphase-'0';
1470 }
1471 if (CDSonly) {
1472 seqstart=CDstart;
1473 seqend=CDend;
1474 if (strand=='-') seqend-=cdsadj;
1475 else seqstart+=cdsadj;
1476 }
1477 else {
1478 seqstart=exons.First()->start;
1479 seqend=exons.Last()->end;
1480 }
1481 int s=0; //resulting nucleotide counter
1482 if (strand=='-') {
1483 for (int x=exons.Count()-1;x>=0;x--) {
1484 uint sgstart=exons[x]->start;
1485 uint sgend=exons[x]->end;
1486 if (seqend<sgstart || seqstart>sgend) continue;
1487 if (seqstart>=sgstart && seqstart<=sgend)
1488 sgstart=seqstart; //seqstart within this segment
1489 if (seqend>=sgstart && seqend<=sgend)
1490 sgend=seqend; //seqend within this segment
1491 if (seglst!=NULL)
1492 seglst->Add(new GSeg(s+1,s+1+sgend-sgstart));
1493 for (uint i=sgend;i>=sgstart;i--) {
1494 spliced[s] = ntComplement(gsubseq[i-start]);
1495 s++;
1496 }//for each nt
1497
1498 if (!CDSonly && cds_start!=NULL && CDstart>0) {
1499 if (CDstart>=sgstart && CDstart<=sgend) {
1500 //CDstart in this segment
1501 //and we are getting the whole transcript
1502 *cds_end=s-(CDstart-sgstart);
1503 }
1504 if (CDend>=sgstart && CDend<=sgend) {
1505 //CDstart in this segment
1506 //and we are getting the whole transcript
1507 *cds_start=s-(CDend-cdsadj-sgstart);
1508 }
1509 }//update local CDS coordinates
1510 } //for each exon
1511 } // - strand
1512 else { // + strand
1513 for (int x=0;x<exons.Count();x++) {
1514 uint sgstart=exons[x]->start;
1515 uint sgend=exons[x]->end;
1516 if (seqend<sgstart || seqstart>sgend) continue;
1517 if (seqstart>=sgstart && seqstart<=sgend)
1518 sgstart=seqstart; //seqstart within this segment
1519 if (seqend>=sgstart && seqend<=sgend)
1520 sgend=seqend; //seqend within this segment
1521 if (seglst!=NULL)
1522 seglst->Add(new GSeg(s+1,s+1+sgend-sgstart));
1523 for (uint i=sgstart;i<=sgend;i++) {
1524 spliced[s]=gsubseq[i-start];
1525 s++;
1526 }//for each nt
1527 if (!CDSonly && cds_start!=NULL && CDstart>0) {
1528 if (CDstart>=sgstart && CDstart<=sgend) {
1529 //CDstart in this segment
1530 //and we are getting the whole transcript
1531 *cds_start=s-(sgend-CDstart-cdsadj);
1532 }
1533 if (CDend>=sgstart && CDend<=sgend) {
1534 //CDstart in this segment
1535 //and we are getting the whole transcript
1536 *cds_end=s-(sgend-CDend);
1537 }
1538 }//update local CDS coordinates
1539 } //for each exon
1540 } // + strand
1541 spliced[s]=0;
1542 if (rlen!=NULL) *rlen=s;
1543 return spliced;
1544 }
1545
1546 char* GffObj::getSplicedTr(GFaSeqGet* faseq, bool CDSonly, int* rlen) {
1547 if (CDSonly && CDstart==0) return NULL;
1548 //restore normal coordinates:
1549 unxcoord();
1550 if (exons.Count()==0) return NULL;
1551 int fspan=end-start+1;
1552 const char* gsubseq=faseq->subseq(start, fspan);
1553 if (gsubseq==NULL) {
1554 GError("Error getting subseq for %s (%d..%d)!\n", gffID, start, end);
1555 }
1556
1557 char* translation=NULL;
1558 GMALLOC(translation, (int)(covlen/3)+1);
1559 uint seqstart, seqend;
1560 int cdsadj=0;
1561 if (CDphase=='1' || CDphase=='2') {
1562 cdsadj=CDphase-'0';
1563 }
1564 if (CDSonly) {
1565 seqstart=CDstart;
1566 seqend=CDend;
1567 if (strand=='-') seqend-=cdsadj;
1568 else seqstart+=cdsadj;
1569 }
1570 else {
1571 seqstart=exons.First()->start;
1572 seqend=exons.Last()->end;
1573 }
1574 Codon codon;
1575 int nt=0; //codon nucleotide counter (0..2)
1576 int aa=0; //aminoacid count
1577 if (strand=='-') {
1578 for (int x=exons.Count()-1;x>=0;x--) {
1579 uint sgstart=exons[x]->start;
1580 uint sgend=exons[x]->end;
1581 if (seqend<sgstart || seqstart>sgend) continue;
1582 if (seqstart>=sgstart && seqstart<=sgend)
1583 sgstart=seqstart; //seqstart within this segment
1584 if (seqend>=sgstart && seqend<=sgend) {
1585 sgend=seqend; //seqend within this segment
1586 }
1587 for (uint i=sgend;i>=sgstart;i--) {
1588 codon.nuc[nt]=ntComplement(gsubseq[i-start]);
1589 nt++;
1590 if (nt==3) {
1591 nt=0;
1592 translation[aa]=codon.translate();
1593 aa++;
1594 }
1595 }//for each nt
1596 } //for each exon
1597 } // - strand
1598 else { // + strand
1599 for (int x=0;x<exons.Count();x++) {
1600 uint sgstart=exons[x]->start;
1601 uint sgend=exons[x]->end;
1602 if (seqend<sgstart || seqstart>sgend) continue;
1603 if (seqstart>=sgstart && seqstart<=sgend)
1604 sgstart=seqstart; //seqstart within this segment
1605 if (seqend>=sgstart && seqend<=sgend)
1606 sgend=seqend; //seqend within this segment
1607 for (uint i=sgstart;i<=sgend;i++) {
1608 codon.nuc[nt]=gsubseq[i-start];
1609 nt++;
1610 if (nt==3) {
1611 nt=0;
1612 translation[aa]=codon.translate();
1613 aa++;
1614 }
1615 }//for each nt
1616 } //for each exon
1617 } // + strand
1618 translation[aa]=0;
1619 if (rlen!=NULL) *rlen=aa;
1620 return translation;
1621 }
1622
1623 void GffObj::printSummary(FILE* fout) {
1624 if (fout==NULL) fout=stdout;
1625 fprintf(fout, "%s\t%c\t%d\t%d\t%4.2f\t%4.1f\n", gffID,
1626 strand, start, end, gscore, (float)qcov/10.0);
1627 }
1628
1629 void GffObj::printGxfLine(FILE* fout, const char* tlabel, const char* gseqname, bool iscds,
1630 uint segstart, uint segend, int exidx, char phase, bool gff3) {
1631 static char scorestr[14];
1632 strcpy(scorestr,".");
1633 GffAttrs* xattrs=NULL;
1634 if (exidx>=0) {
1635 if (exons[exidx]->score) sprintf(scorestr,"%.2f", exons[exidx]->score);
1636 xattrs=exons[exidx]->attrs;
1637 }
1638 if (phase==0 || !iscds) phase='.';
1639 const char* ftype=iscds ? "CDS" : getSubfName();
1640 if (gff3) {
1641 fprintf(fout,
1642 "%s\t%s\t%s\t%d\t%d\t%s\t%c\t%c\tParent=%s",
1643 gseqname, tlabel, ftype, segstart, segend, scorestr, strand,
1644 phase, gffID);
1645 if (xattrs!=NULL) {
1646 for (int i=0;i<xattrs->Count();i++)
1647 fprintf(fout, ";%s=%s",names->attrs.getName(xattrs->Get(i)->attr_id),
1648 xattrs->Get(i)->attr_val);
1649 }
1650 fprintf(fout, "\n");
1651 } //GFF
1652 else {//for GTF -- we print only transcripts
1653 //if (isValidTranscript())
1654 fprintf(fout, "%s\t%s\t%s\t%d\t%d\t%s\t%c\t%c\ttranscript_id \"%s\";",
1655 gseqname, tlabel, ftype, segstart, segend, scorestr, strand, phase, gffID);
1656 //char* geneid=(geneID!=NULL)? geneID : gffID;
1657 if (geneID)
1658 fprintf(fout," gene_id \"%s\";",geneID);
1659 if (gene_name!=NULL) {
1660 //fprintf(fout, " gene_name ");
1661 //if (gene_name[0]=='"') fprintf (fout, "%s;",gene_name);
1662 // else fprintf(fout, "\"%s\";",gene_name);
1663 fprintf(fout," gene_name \"%s\";",gene_name);
1664 }
1665 if (xattrs!=NULL) {
1666 for (int i=0;i<xattrs->Count();i++) {
1667 if (xattrs->Get(i)->attr_val==NULL) continue;
1668 const char* attrname=names->attrs.getName(xattrs->Get(i)->attr_id);
1669 fprintf(fout, " %s ",attrname);
1670 if (xattrs->Get(i)->attr_val[0]=='"')
1671 fprintf(fout, "%s;",xattrs->Get(i)->attr_val);
1672 else fprintf(fout, "\"%s\";",xattrs->Get(i)->attr_val);
1673 }
1674 }
1675 //for GTF, also append the GffObj attributes to each exon line
1676 if ((xattrs=this->attrs)!=NULL) {
1677 for (int i=0;i<xattrs->Count();i++) {
1678 if (xattrs->Get(i)->attr_val==NULL) continue;
1679 const char* attrname=names->attrs.getName(xattrs->Get(i)->attr_id);
1680 fprintf(fout, " %s ",attrname);
1681 if (xattrs->Get(i)->attr_val[0]=='"')
1682 fprintf(fout, "%s;",xattrs->Get(i)->attr_val);
1683 else fprintf(fout, "\"%s\";",xattrs->Get(i)->attr_val);
1684 }
1685 }
1686 fprintf(fout, "\n");
1687 }//GTF
1688 }
1689
1690 void GffObj::printGxf(FILE* fout, GffPrintMode gffp,
1691 const char* tlabel, const char* gfparent) {
1692 static char tmpstr[255];
1693 if (tlabel==NULL) {
1694 tlabel=track_id>=0 ? names->tracks.Get(track_id)->name :
1695 (char*)"gffobj" ;
1696 }
1697 unxcoord();
1698 //if (exons.Count()==0) return;
1699 const char* gseqname=names->gseqs.Get(gseq_id)->name;
1700 bool gff3 = (gffp>=pgffAny);
1701 bool showCDS = (gffp==pgtfAny || gffp==pgtfCDS || gffp==pgffCDS || gffp==pgffAny || gffp==pgffBoth);
1702 bool showExon = (gffp<=pgtfExon || gffp==pgffAny || gffp==pgffExon || gffp==pgffBoth);
1703 if (gff3) {
1704 //print GFF3 mRNA line:
1705 if (gscore>0.0) sprintf(tmpstr,"%.2f", gscore);
1706 else strcpy(tmpstr,".");
1707 uint pstart, pend;
1708 if (gffp==pgffCDS) {
1709 pstart=CDstart;
1710 pend=CDend;
1711 }
1712 else { pstart=start;pend=end; }
1713 //const char* ftype=isTranscript() ? "mRNA" : getFeatureName();
1714 const char* ftype=getFeatureName();
1715 fprintf(fout,
1716 "%s\t%s\t%s\t%d\t%d\t%s\t%c\t.\tID=%s",
1717 gseqname, tlabel, ftype, pstart, pend, tmpstr, strand, gffID);
1718 if (CDstart>0 && !showCDS && !isCDS) fprintf(fout,";CDS=%d-%d",CDstart,CDend);
1719 if (gfparent!=NULL) {
1720 //parent override
1721 fprintf(fout, ";Parent=%s",gfparent);
1722 }
1723 else {
1724 if (parent!=NULL && !parent->isDiscarded())
1725 fprintf(fout, ";Parent=%s",parent->getID());
1726 }
1727 if (geneID!=NULL)
1728 fprintf(fout, ";geneID=%s",geneID);
1729 if (gene_name!=NULL)
1730 fprintf(fout, ";gene_name=%s",gene_name);
1731 if (attrs!=NULL) {
1732 for (int i=0;i<attrs->Count();i++) {
1733 const char* attrname=names->attrs.getName(attrs->Get(i)->attr_id);
1734 fprintf(fout,";%s=%s", attrname,
1735 attrs->Get(i)->attr_val);
1736 }
1737 }
1738 fprintf(fout,"\n");
1739 }// gff3 mRNA line
1740 if (showExon) {
1741 //print exons
1742 if (isCDS && exons.Count()>0 &&
1743 ((strand=='-' && exons.Last()->phase<'0') || (strand=='+' && exons.Last()->phase<'0')))
1744 updateExonPhase();
1745
1746 for (int i=0;i<exons.Count();i++) {
1747 printGxfLine(fout, tlabel, gseqname, isCDS, exons[i]->start, exons[i]->end, i, exons[i]->phase, gff3);
1748 }
1749 }//printing exons
1750 if (showCDS && !isCDS && CDstart>0) {
1751 GArray<GffCDSeg> cds(true,true);
1752 getCDSegs(cds);
1753 for (int i=0;i<cds.Count();i++) {
1754 printGxfLine(fout, tlabel, gseqname, true, cds[i].start, cds[i].end, -1, cds[i].phase, gff3);
1755 }
1756 } //showCDS
1757 }
1758
1759 void GffObj::updateExonPhase() {
1760 if (!isCDS) return;
1761 int cdsacc=0;
1762 if (CDphase=='1' || CDphase=='2') {
1763 cdsacc+= 3-(CDphase-'0');
1764 }
1765 if (strand=='-') { //reverse strand
1766 for (int i=exons.Count()-1;i>=0;i--) {
1767 exons[i]->phase='0'+ (3-cdsacc%3)%3;
1768 cdsacc+=exons[i]->end-exons[i]->start+1;
1769 }
1770 }
1771 else { //forward strand
1772 for (int i=0;i<exons.Count();i++) {
1773 exons[i]->phase='0'+ (3-cdsacc%3)%3;
1774 cdsacc+=exons[i]->end-exons[i]->start+1;
1775 }
1776 }
1777 }
1778
1779
1780 void GffObj::getCDSegs(GArray<GffCDSeg>& cds) {
1781 GffCDSeg cdseg;
1782 int cdsacc=0;
1783 if (CDphase=='1' || CDphase=='2') {
1784 cdsacc+= 3-(CDphase-'0');
1785 }
1786 if (strand=='-') {
1787 for (int x=exons.Count()-1;x>=0;x--) {
1788 uint sgstart=exons[x]->start;
1789 uint sgend=exons[x]->end;
1790 if (CDend<sgstart || CDstart>sgend) continue;
1791 if (CDstart>=sgstart && CDstart<=sgend)
1792 sgstart=CDstart; //cdstart within this segment
1793 if (CDend>=sgstart && CDend<=sgend)
1794 sgend=CDend; //cdend within this segment
1795 cdseg.start=sgstart;
1796 cdseg.end=sgend;
1797 cdseg.exonidx=x;
1798 //cdseg.phase='0'+(cdsacc>0 ? (3-cdsacc%3)%3 : 0);
1799 cdseg.phase='0'+ (3-cdsacc%3)%3;
1800 cdsacc+=sgend-sgstart+1;
1801 cds.Add(cdseg);
1802 } //for each exon
1803 } // - strand
1804 else { // + strand
1805 for (int x=0;x<exons.Count();x++) {
1806 uint sgstart=exons[x]->start;
1807 uint sgend=exons[x]->end;
1808 if (CDend<sgstart || CDstart>sgend) continue;
1809 if (CDstart>=sgstart && CDstart<=sgend)
1810 sgstart=CDstart; //seqstart within this segment
1811 if (CDend>=sgstart && CDend<=sgend)
1812 sgend=CDend; //seqend within this segment
1813 cdseg.start=sgstart;
1814 cdseg.end=sgend;
1815 cdseg.exonidx=x;
1816 //cdseg.phase='0'+(cdsacc>0 ? (3-cdsacc%3)%3 : 0);
1817 cdseg.phase='0' + (3-cdsacc%3)%3 ;
1818 cdsacc+=sgend-sgstart+1;
1819 cds.Add(cdseg);
1820 } //for each exon
1821 } // + strand
1822 }