ViewVC Help
View File | Revision Log | Show Annotations | View Changeset | Root Listing
root/gclib/gclib/gff.cpp
Revision: 51
Committed: Thu Sep 8 01:02:11 2011 UTC (7 years, 10 months ago) by gpertea
File size: 64797 byte(s)
Log Message:
fix direct CDS segment given as the only subfeature

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 /*
576 if (start>0 && exontype==exgffCDS && exons.Count()==0) {
577 //adding a CDS directly as the first subfeature of a declared parent
578 segstart=start;
579 segend=end;
580 }
581 */
582 GffExon* enew=new GffExon(segstart, segend, sc, fr, qs, qe, exontype);
583 int eidx=exons.Add(enew);
584 if (eidx<0) {
585 //this would actually be acceptable if the object is a "Gene" and "exons" are in fact isoforms
586 if (gff_show_warnings)
587 GMessage("GFF Warning: failed adding segment %d-%d for %s (discarded)!\n",
588 segstart, segend, gffID);
589 delete enew;
590 hasErrors(true);
591 return -1;
592 }
593 covlen+=(int)(exons[eidx]->end-exons[eidx]->start)+1;
594 if (start==0) {
595 start=exons.First()->start;
596 end=exons.Last()->end;
597 }
598 if (uptr!=NULL) { //collect stats about the underlying genomic sequence
599 GSeqStat* gsd=(GSeqStat*)uptr;
600 if (start<gsd->mincoord) gsd->mincoord=start;
601 if (end>gsd->maxcoord) gsd->maxcoord=end;
602 if (this->len()>gsd->maxfeat_len) {
603 gsd->maxfeat_len=this->len();
604 gsd->maxfeat=this;
605 }
606 }
607 return eidx;
608 }
609
610 void GffObj::expandExon(int oi, uint segstart, uint segend, char exontype, double sc, char fr, int qs, int qe) {
611 //oi is the index of the *first* overlapping segment found that must be enlarged
612 covlen-=exons[oi]->len();
613 if (segstart<exons[oi]->start)
614 exons[oi]->start=segstart;
615 if (qs && qs<exons[oi]->qstart) exons[oi]->qstart=qs;
616 if (segend>exons[oi]->end)
617 exons[oi]->end=segend;
618 if (qe && qe>exons[oi]->qend) exons[oi]->qend=qe;
619 //warning: score cannot be properly adjusted! e.g. if it's a p-value it's just going to get worse
620 if (sc!=0) exons[oi]->score=sc;
621 covlen+=exons[oi]->len();
622 //if (exons[oi]->exontype< exontype) -- always true
623 exons[oi]->exontype = exontype;
624 if (exontype==exgffCDS) exons[oi]->phase=fr;
625 //we must check if any more exons are also overlapping this
626 int ni=oi+1; //next exon index after oi
627 while (ni<exons.Count() && segend>=exons[ni]->start) { // next segment overlaps new enlarged segment
628 //only allow this if next segment is fully included, and a subordinate
629 if (exons[ni]->exontype<exontype && exons[ni]->end<=segend) {
630 /* I guess we have to relax this due to stupid UCSC hg18 files having a start_codon sticking out
631 chr1 hg18_knownGene start_codon 69806911 69806913 0.000000 + .
632 chr1 hg18_knownGene CDS 69806911 69806912 0.000000 + 0
633 chr1 hg18_knownGene exon 69805456 69806912 0.000000 + .
634 */
635 if (exons[ni]->qstart<exons[oi]->qstart) exons[oi]->qstart=exons[ni]->qstart;
636 if (exons[ni]->qend>exons[oi]->qend) exons[oi]->qend=exons[ni]->qend;
637 exons.Delete(ni);
638 }
639 else {
640 if (gff_show_warnings) GMessage("GFF Warning: overlapping existing exon(%d-%d) while expanding to %d-%d for GFF ID %s\n",
641 exons[ni]->start, exons[ni]->end, segstart, segend, gffID);
642 //hasErrors(true);
643 break;
644 }
645 }
646 // -- make sure any other related boundaries are updated:
647 start=exons.First()->start;
648 end=exons.Last()->end;
649 if (uptr!=NULL) { //collect stats about the underlying genomic sequence
650 GSeqStat* gsd=(GSeqStat*)uptr;
651 if (start<gsd->mincoord) gsd->mincoord=start;
652 if (end>gsd->maxcoord) gsd->maxcoord=end;
653 if (this->len()>gsd->maxfeat_len) {
654 gsd->maxfeat_len=this->len();
655 gsd->maxfeat=this;
656 }
657 }
658 }
659
660 void GffObj::removeExon(int idx) {
661 /*
662 if (idx==0 && segs[0].start==gstart)
663 gstart=segs[1].start;
664 if (idx==segcount && segs[segcount].end==gend)
665 gend=segs[segcount-1].end;
666 */
667 if (idx<0 || idx>=exons.Count()) return;
668 int segstart=exons[idx]->start;
669 int segend=exons[idx]->end;
670 exons.Delete(idx);
671 covlen -= (int)(segend-segstart)+1;
672 start=exons.First()->start;
673 end=exons.Last()->end;
674 if (isCDS) { CDstart=start; CDend=end; }
675 }
676
677 void GffObj::removeExon(GffExon* p) {
678 for (int idx=0;idx<exons.Count();idx++) {
679 if (exons[idx]==p) {
680 int segstart=exons[idx]->start;
681 int segend=exons[idx]->end;
682 exons.Delete(idx);
683 covlen -= (int)(segend-segstart)+1;
684 start=exons.First()->start;
685 end=exons.Last()->end;
686 if (isCDS) { CDstart=start; CDend=end; }
687 return;
688 }
689 }
690 }
691
692
693
694 GffObj::GffObj(GffReader *gfrd, GffLine* gffline, bool keepAttr, bool noExonAttr):
695 GSeg(0,0), exons(true,true,false), children(1,false) {
696 xstart=0;
697 xend=0;
698 xstatus=0;
699 partial=false;
700 isCDS=false;
701 uptr=NULL;
702 ulink=NULL;
703 parent=NULL;
704 udata=0;
705 flags=0;
706 CDstart=0;
707 CDend=0;
708 CDphase=0;
709 geneID=NULL;
710 gene_name=NULL;
711 attrs=NULL;
712 gffID=NULL;
713 track_id=-1;
714 gseq_id=-1;
715 ftype_id=-1;
716 exon_ftype_id=-1;
717 strand='.';
718 if (gfrd==NULL)
719 GError("Cannot use this GffObj constructor with a NULL GffReader!\n");
720 gffnames_ref(names);
721 if (gfrd->names==NULL) gfrd->names=names;
722 //qlen=0;qstart=0;qend=0;
723 gscore=0;
724 uscore=0;
725 covlen=0;
726 qcov=0;
727 start=gffline->fstart;
728 end=gffline->fend;
729 gseq_id=names->gseqs.addName(gffline->gseqname);
730 track_id=names->tracks.addName(gffline->track);
731 strand=gffline->strand;
732 qlen=gffline->qlen;
733 qstart=gffline->qstart;
734 qend=gffline->qend;
735 //setup flags from gffline
736 isCDS=gffline->is_cds; //for now
737 isGene(gffline->is_gene);
738 isTranscript(gffline->is_transcript || gffline->exontype!=0);
739 fromGff3(gffline->is_gff3);
740
741 if (gffline->parents!=NULL) {
742 //GTF style -- create a GffObj directly by subfeature
743 //(also possible orphan GFF3 exon line, or an exon given before its parent (chado))
744 if (gffline->exontype!=0) { //recognized exon-like feature
745 ftype_id=gff_fid_transcript; //so this is some sort of transcript
746 exon_ftype_id=gff_fid_exon; //subfeatures MUST be exons
747 }
748 else {//unrecognized subfeatures
749 //make this GffObj of the same feature type
750 ftype_id=names->feats.addName(gffline->ftype);
751 }
752 if (gffline->ID==NULL) { //typical GTF
753 gffID=Gstrdup(gffline->parents[0]);
754 this->createdByExon(true);
755 //this is likely the first exon/segment of the feature
756 addExon(gfrd, gffline, keepAttr, noExonAttr);
757 }
758 else { //a parented feature with an ID -- probably an orphan GFF3 line
759 if (gffline->is_gff3 && gffline->exontype!=0) {
760 //premature exon given before its parent transcript
761 //create the transcript entry here
762 gffID=Gstrdup(gffline->parents[0]);
763 this->createdByExon(true);
764 //this is the first exon/segment of the transcript
765 addExon(gfrd, gffline, keepAttr, noExonAttr);
766 }
767 else { //unrecognized non-exon feature ? use the ID instead
768 gffID=Gstrdup(gffline->ID);
769 if (keepAttr) this->parseAttrs(attrs, gffline->info);
770 }
771 }
772 } //subfeature given directly
773 else { //gffline->parents==NULL
774 //create a parent feature in its own right
775 gscore=gffline->score;
776 if (gffline->ID==NULL || gffline->ID[0]==0)
777 GError("Error: no ID found for GFF record start\n");
778 gffID=Gstrdup(gffline->ID); //there must be an ID here
779 //if (gffline->is_transcript) ftype_id=gff_fid_mRNA;
780 //else
781 ftype_id=names->feats.addName(gffline->ftype);
782 if (gffline->is_transcript)
783 exon_ftype_id=gff_fid_exon;
784
785 if (keepAttr) this->parseAttrs(attrs, gffline->info);
786 }//no parent
787
788 if (gffline->gene_name!=NULL) {
789 gene_name=Gstrdup(gffline->gene_name);
790 }
791 if (gffline->gene_id!=NULL) {
792 geneID=Gstrdup(gffline->gene_id);
793 }
794
795 GSeqStat* gsd=gfrd->gseqstats.AddIfNew(new GSeqStat(gseq_id,names->gseqs.lastNameUsed()),true);
796 uptr=gsd;
797 if (start<gsd->mincoord) gsd->mincoord=start;
798 if (end>gsd->maxcoord) gsd->maxcoord=end;
799 if (this->len()>gsd->maxfeat_len) {
800 gsd->maxfeat_len=this->len();
801 gsd->maxfeat=this;
802 }
803 }
804
805 GffLine* GffReader::nextGffLine() {
806 if (gffline!=NULL) return gffline; //caller should free gffline after processing
807 while (gffline==NULL) {
808 int llen=0;
809 buflen=GFF_LINELEN-1;
810 char* l=fgetline(linebuf, buflen, fh, &fpos, &llen);
811 if (l==NULL) {
812 return NULL; //end of file
813 }
814 int ns=0; //first nonspace position
815 while (l[ns]!=0 && isspace(l[ns])) ns++;
816 if (l[ns]=='#' || llen<10) continue;
817 gffline=new GffLine(this, l);
818 if (gffline->skip) {
819 delete gffline;
820 gffline=NULL;
821 continue;
822 }
823 if (gffline->ID==NULL && gffline->parents==NULL) { //it must have an ID
824 //this might not be needed, already checked in the GffLine constructor
825 if (gff_warns)
826 GMessage("Warning: malformed GFF line, no parent or record Id (kipping\n");
827 delete gffline;
828 gffline=NULL;
829 //continue;
830 }
831 }
832 return gffline;
833 }
834
835 char* GffReader::gfoBuildId(const char* id, const char* ctg) {
836 //caller must free the returned pointer
837 char* buf=NULL;
838 int idlen=strlen(id);
839 GMALLOC(buf, idlen+strlen(ctg)+2);
840 strcpy(buf, id);
841 buf[idlen]='~';
842 strcpy(buf+idlen+1, ctg);
843 return buf;
844 }
845
846 void GffReader::gfoRemove(const char* id, const char* ctg) {
847 char* buf=gfoBuildId(id,ctg);
848 phash.Remove(buf);
849 GFREE(buf);
850 }
851
852 //Warning: if gflst gets altered, idx becomes obsolete
853 GfoHolder* GffReader::gfoAdd(const char* id, const char* ctg, GffObj* gfo, int idx) {
854 char* buf=gfoBuildId(id,ctg);
855 GfoHolder* r=new GfoHolder(gfo,idx);
856 phash.Add(buf, r);
857 GFREE(buf);
858 return r;
859 }
860
861 GfoHolder* GffReader::gfoFind(const char* id, const char* ctg) {
862 char* buf=gfoBuildId(id,ctg);
863 GfoHolder* r=phash.Find(buf);
864 GFREE(buf);
865 return r;
866 }
867
868 GfoHolder* GffReader::replaceGffRec(GffLine* gffline, bool keepAttr, bool noExonAttr, int replaceidx) {
869 GffObj* newgfo=new GffObj(this, gffline, keepAttr, noExonAttr);
870 GfoHolder* r=NULL;
871 if (replaceidx>=0) {
872 gflst.Put(replaceidx,newgfo);
873 r=gfoAdd(newgfo->gffID, gffline->gseqname, newgfo, replaceidx);
874 }
875 else {
876 int gfoidx=gflst.Add(newgfo);
877 r=gfoAdd(newgfo->gffID, gffline->gseqname, newgfo, gfoidx);
878 }
879 if (gff_warns) {
880 int* pcount=tids.Find(newgfo->gffID);
881 if (pcount!=NULL) {
882 if (gff_warns) GMessage("Warning: duplicate GFF ID: %s\n", newgfo->gffID);
883 (*pcount)++;
884 }
885 else {
886 tids.Add(newgfo->gffID,new int(1));
887 }
888 }
889 return r;
890 }
891
892 GfoHolder* GffReader::updateParent(GfoHolder* newgfh, GffObj* parent) {
893 //assert(parent);
894 //assert(newgfo);
895 parent->children.Add(newgfh->gffobj);
896 if (newgfh->gffobj->parent==NULL) newgfh->gffobj->parent=parent;
897 newgfh->gffobj->setLevel(parent->getLevel()+1);
898 if (parent->isGene()) {
899 if (parent->gene_name!=NULL && newgfh->gffobj->gene_name==NULL)
900 newgfh->gffobj->gene_name=Gstrdup(parent->gene_name);
901 if (parent->geneID!=NULL && newgfh->gffobj->geneID==NULL)
902 newgfh->gffobj->geneID=Gstrdup(parent->geneID);
903 }
904
905 return newgfh;
906 }
907
908 GfoHolder* GffReader::newGffRec(GffLine* gffline, bool keepAttr, bool noExonAttr,
909 GffObj* parent, GffExon* pexon) {
910 GffObj* newgfo=new GffObj(this, gffline, keepAttr, noExonAttr);
911 GfoHolder* r=NULL;
912 int gfoidx=gflst.Add(newgfo);
913 r=gfoAdd(newgfo->gffID, gffline->gseqname, newgfo, gfoidx);
914 if (parent!=NULL) {
915 updateParent(r, parent);
916 if (pexon!=NULL) parent->removeExon(pexon);
917 }
918 if (gff_warns) {
919 int* pcount=tids.Find(newgfo->gffID);
920 if (pcount!=NULL) {
921 if (gff_warns) GMessage("Warning: duplicate GFF ID: %s\n", newgfo->gffID);
922 (*pcount)++;
923 }
924 else {
925 tids.Add(newgfo->gffID,new int(1));
926 }
927 }
928 return r;
929 }
930
931 GfoHolder* GffReader::updateGffRec(GfoHolder* prevgfo, GffLine* gffline,
932 bool keepAttr) {
933 if (prevgfo==NULL) return NULL;
934 prevgfo->gffobj->createdByExon(false);
935 prevgfo->gffobj->ftype_id=prevgfo->gffobj->names->feats.addName(gffline->ftype);
936 prevgfo->gffobj->start=gffline->fstart;
937 prevgfo->gffobj->end=gffline->fend;
938 prevgfo->gffobj->isGene(gffline->is_gene);
939 prevgfo->gffobj->isTranscript(gffline->is_transcript || gffline->exontype!=0);
940 prevgfo->gffobj->fromGff3(gffline->is_gff3);
941 if (keepAttr) {
942 if (prevgfo->gffobj->attrs!=NULL) prevgfo->gffobj->attrs->Clear();
943 prevgfo->gffobj->parseAttrs(prevgfo->gffobj->attrs, gffline->info);
944 }
945 return prevgfo;
946 }
947
948
949 bool GffReader::addExonFeature(GfoHolder* prevgfo, GffLine* gffline, GHash<CNonExon>& pex, bool noExonAttr) {
950 bool r=true;
951 if (gffline->strand!=prevgfo->gffobj->strand) {
952 GMessage("GFF Error: duplicate GFF ID '%s' (exons found on different strands of %s)\n",
953 prevgfo->gffobj->gffID, prevgfo->gffobj->getGSeqName());
954 r=false;
955 }
956 int gdist=(gffline->fstart>prevgfo->gffobj->end) ? gffline->fstart-prevgfo->gffobj->end :
957 ((gffline->fend<prevgfo->gffobj->start)? prevgfo->gffobj->start-gffline->fend :
958 0 );
959 if (gdist>(int)GFF_MAX_LOCUS) { //too far apart, most likely this is a duplicate ID
960 GMessage("Error: duplicate GFF ID '%s' (or exons too far apart)!\n",prevgfo->gffobj->gffID);
961 //validation_errors = true;
962 r=false;
963 if (!gff_warns) exit(1);
964 }
965 int eidx=prevgfo->gffobj->addExon(this, gffline, !noExonAttr, noExonAttr);
966 if (eidx>=0 && gffline->ID!=NULL && gffline->exontype==0)
967 subfPoolAdd(pex, prevgfo);
968 return r;
969 }
970
971 CNonExon* GffReader::subfPoolCheck(GffLine* gffline, GHash<CNonExon>& pex, char*& subp_name) {
972 CNonExon* subp=NULL;
973 subp_name=NULL;
974 for (int i=0;i<gffline->num_parents;i++) {
975 if (transcriptsOnly && discarded_ids.Find(gffline->parents[i])!=NULL)
976 continue;
977 subp_name=gfoBuildId(gffline->parents[i], gffline->gseqname); //e.g. mRNA name
978 subp=pex.Find(subp_name);
979 if (subp!=NULL)
980 return subp;
981 GFREE(subp_name);
982 }
983 return NULL;
984 }
985
986 void GffReader::subfPoolAdd(GHash<CNonExon>& pex, GfoHolder* newgfo) {
987 //this might become a parent feature later
988 if (newgfo->gffobj->exons.Count()>0) {
989 char* xbuf=gfoBuildId(gffline->ID, gffline->gseqname);
990 pex.Add(xbuf, new CNonExon(newgfo->idx, newgfo->gffobj,
991 newgfo->gffobj->exons[0], gffline));
992 GFREE(xbuf);
993 }
994 }
995
996 GfoHolder* GffReader::promoteFeature(CNonExon* subp, char*& subp_name, GHash<CNonExon>& pex,
997 bool keepAttr, bool noExonAttr) {
998 GffObj* prevp=subp->parent; //grandparent of gffline (e.g. gene)
999 if (prevp!=gflst[subp->idx])
1000 GError("Error promoting subfeature %s, gflst index mismatch?!\n", subp->gffline->ID);
1001 subp->gffline->discardParent();
1002 GfoHolder* gfoh=newGffRec(subp->gffline, keepAttr, noExonAttr, prevp, subp->exon);
1003 pex.Remove(subp_name); //no longer a potential parent, moved it to phash already
1004 prevp->promotedChildren(true);
1005 return gfoh; //returns the holder of newly promoted feature
1006 }
1007
1008 //have to parse the whole file because exons can be scattered all over
1009 void GffReader::readAll(bool keepAttr, bool mergeCloseExons, bool noExonAttr) {
1010 bool validation_errors = false;
1011 //loc_debug=false;
1012 GHash<CNonExon> pex; //keep track of any "exon"-like features that have an ID
1013 //and thus could become promoted to parent features
1014 while (nextGffLine()!=NULL) {
1015 //seen this gff ID before?
1016 GfoHolder* prevseen=NULL;
1017 if (gffline->ID) //GFF3
1018 prevseen=gfoFind(gffline->ID, gffline->gseqname);
1019 if (prevseen!=NULL) {
1020 if (prevseen->gffobj->createdByExon()) {
1021 updateGffRec(prevseen, gffline, keepAttr);
1022 }
1023 else {
1024 GMessage("Error: duplicate GFF ID '%s' encountered!\n",gffline->ID);
1025 validation_errors = true;
1026 if (gff_warns) {
1027 delete gffline; gffline=NULL; continue;
1028 }
1029 else exit(1);
1030 }
1031 }
1032 if (gffline->parents==NULL) {//start GFF3-like record with no parent (mRNA, gene)
1033 if (!prevseen) newGffRec(gffline, keepAttr, noExonAttr);
1034 }
1035 else { //--- it's a parented feature (could still be a mRNA)
1036 bool found_parent=false;
1037 GfoHolder* newgfo=prevseen;
1038 for (int i=0;i<gffline->num_parents;i++) {
1039 if (transcriptsOnly && discarded_ids.Find(gffline->parents[i])!=NULL)
1040 continue; //skipping discarded parent feature
1041 GfoHolder* parentgfo=gfoFind(gffline->parents[i], gffline->gseqname);
1042 if (parentgfo!=NULL) { //parent GffObj parsed earlier
1043 found_parent=true;
1044 if (parentgfo->gffobj->isGene() && gffline->is_transcript
1045 && gffline->exontype==0) {
1046 //not an exon, but a transcript parented by a gene
1047 if (newgfo) {
1048 updateParent(newgfo, parentgfo->gffobj);
1049 }
1050 else {
1051 newgfo=newGffRec(gffline, keepAttr, noExonAttr, parentgfo->gffobj);
1052 }
1053 }
1054 else { //potential exon subfeature
1055 if (!addExonFeature(parentgfo, gffline, pex, noExonAttr))
1056 validation_errors=true;
1057 }
1058 }
1059 } //for each parsed parent Id
1060 if (!found_parent) { //new GTF-like record starting here with a subfeature directly
1061 //or it could be some chado GFF3 barf with exons declared BEFORE their parent :(
1062 //check if this feature isn't parented by a previously stored "exon" subfeature
1063 char* subp_name=NULL;
1064 CNonExon* subp=subfPoolCheck(gffline, pex, subp_name);
1065 if (subp!=NULL) { //found a subfeature that is the parent of this gffline
1066 //promote that subfeature to a full GffObj
1067 GfoHolder* gfoh=promoteFeature(subp, subp_name, pex, keepAttr, noExonAttr);
1068 //add current gffline as an exon of the newly promoted subfeature
1069 if (!addExonFeature(gfoh, gffline, pex, noExonAttr))
1070 validation_errors=true;
1071 }
1072 else { //no parent seen before, create one directly with this exon
1073 //loc_debug=true;
1074 GfoHolder* newgfo=prevseen ? prevseen : newGffRec(gffline, keepAttr, noExonAttr);
1075 if (gffline->ID!=NULL && gffline->exontype==0)
1076 subfPoolAdd(pex, newgfo);
1077 //even those with errors will be added here!
1078 }
1079 GFREE(subp_name);
1080 } //no previous parent found
1081 } //parented feature
1082 //--
1083 delete gffline;
1084 gffline=NULL;
1085 }//while gff lines
1086 gflst.finalize(this, mergeCloseExons, keepAttr, noExonAttr); //force sorting by locus if so constructed
1087 // all gff records are now loaded in GList gflst
1088 // so we can free the hash
1089 phash.Clear();
1090 tids.Clear();
1091 if (validation_errors) {
1092 exit(1);
1093 }
1094 }
1095
1096 GffObj* GffObj::finalize(GffReader* gfr, bool mergeCloseExons, bool keepAttrs, bool noExonAttr) {
1097 //merge
1098 //always merge adjacent or overlapping segments
1099 //but if mergeCloseExons then merge even when distance is up to 5 bases
1100 udata=0;
1101 uptr=NULL;
1102 if (gfr->transcriptsOnly && !(isTranscript() || (isGene() && children.Count()==0))) {
1103 isDiscarded(true);
1104 }
1105 if (ftype_id==gff_fid_transcript && CDstart>0) {
1106 ftype_id=gff_fid_mRNA;
1107 //exon_ftype_id=gff_fid_exon;
1108 }
1109 //if (ftype_id==gff_fid_mRNA || exon_ftype_id==gff_fid_exon || mergeCloseExons) {
1110 if (isTranscript() || exon_ftype_id==gff_fid_exon || mergeCloseExons) {
1111 int mindist=mergeCloseExons ? 5:1;
1112 for (int i=0;i<exons.Count()-1;i++) {
1113 int ni=i+1;
1114 uint mend=exons[i]->end;
1115 while (ni<exons.Count()) {
1116 int dist=(int)(exons[ni]->start-mend);
1117 if (dist>mindist) break; //no merging with next segment
1118 if (gfr!=NULL && gfr->gff_warns && dist!=0 && (exons[ni]->exontype!=exgffUTR && exons[i]->exontype!=exgffUTR)) {
1119 GMessage("GFF warning: merging adjacent/overlapping segments of %s on %s (%d-%d, %d-%d)\n",
1120 gffID, getGSeqName(), exons[i]->start, exons[i]->end,exons[ni]->start, exons[ni]->end);
1121 }
1122 mend=exons[ni]->end;
1123 covlen-=exons[i]->len();
1124 exons[i]->end=mend;
1125 covlen+=exons[i]->len();
1126 covlen-=exons[ni]->len();
1127 if (exons[ni]->attrs!=NULL && (exons[i]->attrs==NULL ||
1128 exons[i]->attrs->Count()<exons[ni]->attrs->Count())) {
1129 //use the other exon attributes, if more
1130 delete(exons[i]->attrs);
1131 exons[i]->attrs=exons[ni]->attrs;
1132 exons[ni]->attrs=NULL;
1133 }
1134 exons.Delete(ni);
1135 } //check for merge with next exon
1136 } //for each exon
1137 }
1138 //attribute reduction for GTF records
1139 if (keepAttrs && !noExonAttr && !fromGff3()
1140 && exons.Count()>0 && exons[0]->attrs!=NULL) {
1141 bool attrs_discarded=false;
1142 for (int a=0;a<exons[0]->attrs->Count();a++) {
1143 int attr_name_id=exons[0]->attrs->Get(a)->attr_id;
1144 char* attr_name=names->attrs.getName(attr_name_id);
1145 char* attr_val =exons[0]->attrs->Get(a)->attr_val;
1146 bool sameExonAttr=true;
1147 for (int i=1;i<exons.Count();i++) {
1148 char* ov=exons[i]->getAttr(attr_name_id);
1149 if (ov==NULL || (strcmp(ov,attr_val)!=0)) {
1150 sameExonAttr=false;
1151 break;
1152 }
1153 }
1154 if (sameExonAttr) {
1155 //delete this attribute from exons level
1156 attrs_discarded=true;
1157 this->addAttr(attr_name, attr_val);
1158 for (int i=1;i<exons.Count();i++) {
1159 removeExonAttr(*(exons[i]), attr_name_id);
1160 }
1161 exons[0]->attrs->freeItem(a);
1162 }
1163 }
1164 if (attrs_discarded) exons[0]->attrs->Pack();
1165 }
1166 return this;
1167 }
1168
1169 void GffObj::parseAttrs(GffAttrs*& atrlist, char* info, bool isExon) {
1170 if (names==NULL)
1171 GError(ERR_NULL_GFNAMES, "parseAttrs()");
1172 if (atrlist==NULL)
1173 atrlist=new GffAttrs();
1174 char* endinfo=info+strlen(info);
1175 char* start=info;
1176 char* pch=start;
1177 while (start<endinfo) {
1178 //skip spaces
1179 while (*start==' ' && start<endinfo) start++;
1180 pch=strchr(start, ';');
1181 if (pch==NULL) pch=endinfo;
1182 else {
1183 *pch='\0';
1184 pch++;
1185 }
1186 char* ech=strchr(start,'=');
1187 if (ech!=NULL) { // attr=value format found
1188 *ech='\0';
1189 //if (noExonAttr && (strcmp(start, "exon_number")==0 || strcmp(start, "exon")==0)) { start=pch; continue; }
1190 if (strcmp(start, "exon_number")==0 || strcmp(start, "exon")==0 ||
1191 strcmp(start, "exon_id")==0)
1192 { start=pch; continue; }
1193 ech++;
1194 while (*ech==' ' && ech<endinfo) ech++;//skip extra spaces after the '='
1195 //atrlist->Add(new GffAttr(names->attrs.addName(start),ech));
1196 //make sure we don't add the same attribute more than once
1197 if (isExon && (strcmp(start, "protein_id")==0)) {
1198 //Ensembl special case
1199 this->addAttr(start, ech);
1200 start=pch;
1201 continue;
1202 }
1203 atrlist->add_or_update(names, start, ech);
1204 }
1205 /*
1206 else { //not an attr=value format
1207 atrlist->Add(new GffAttr(names->attrs.addName(start),"1"));
1208 }
1209 */
1210 start=pch;
1211 }
1212 if (atrlist->Count()==0) { delete atrlist; atrlist=NULL; }
1213 }
1214
1215 void GffObj::addAttr(const char* attrname, const char* attrvalue) {
1216 if (this->attrs==NULL)
1217 this->attrs=new GffAttrs();
1218 //this->attrs->Add(new GffAttr(names->attrs.addName(attrname),attrvalue));
1219 this->attrs->add_or_update(names, attrname, attrvalue);
1220 }
1221
1222
1223 void GffObj::setFeatureName(const char* feature) {
1224 //change the feature name/type for a transcript
1225 int fid=names->feats.addName(feature);
1226 if (monoFeature() && exons.Count()>0)
1227 this->exon_ftype_id=fid;
1228 this->ftype_id=fid;
1229 }
1230
1231 void GffObj::setRefName(const char* newname) {
1232 //change the feature name/type for a transcript
1233 int rid=names->gseqs.addName(newname);
1234 this->gseq_id=rid;
1235 }
1236
1237
1238
1239 int GffObj::removeAttr(const char* attrname, const char* attrval) {
1240 if (this->attrs==NULL || attrname==NULL || attrname[0]==0) return 0;
1241 int aid=this->names->attrs.getId(attrname);
1242 if (aid<0) return 0;
1243 int delcount=0; //could be more than one ?
1244 for (int i=0;i<this->attrs->Count();i++) {
1245 if (aid==this->attrs->Get(i)->attr_id) {
1246 if (attrval==NULL ||
1247 strcmp(attrval, this->attrs->Get(i)->attr_val)==0) {
1248 delcount++;
1249 this->attrs->freeItem(i);
1250 }
1251 }
1252 }
1253 if (delcount>0) this->attrs->Pack();
1254 return delcount;
1255 }
1256
1257 int GffObj::removeAttr(int aid, const char* attrval) {
1258 if (this->attrs==NULL || aid<0) return 0;
1259 int delcount=0; //could be more than one ?
1260 for (int i=0;i<this->attrs->Count();i++) {
1261 if (aid==this->attrs->Get(i)->attr_id) {
1262 if (attrval==NULL ||
1263 strcmp(attrval, this->attrs->Get(i)->attr_val)==0) {
1264 delcount++;
1265 this->attrs->freeItem(i);
1266 }
1267 }
1268 }
1269 if (delcount>0) this->attrs->Pack();
1270 return delcount;
1271 }
1272
1273
1274 int GffObj::removeExonAttr(GffExon& exon, const char* attrname, const char* attrval) {
1275 if (exon.attrs==NULL || attrname==NULL || attrname[0]==0) return 0;
1276 int aid=this->names->attrs.getId(attrname);
1277 if (aid<0) return 0;
1278 int delcount=0; //could be more than one
1279 for (int i=0;i<exon.attrs->Count();i++) {
1280 if (aid==exon.attrs->Get(i)->attr_id) {
1281 if (attrval==NULL ||
1282 strcmp(attrval, exon.attrs->Get(i)->attr_val)==0) {
1283 delcount++;
1284 exon.attrs->freeItem(i);
1285 }
1286 }
1287 }
1288 if (delcount>0) exon.attrs->Pack();
1289 return delcount;
1290 }
1291
1292 int GffObj::removeExonAttr(GffExon& exon, int aid, const char* attrval) {
1293 if (exon.attrs==NULL || aid<0) return 0;
1294 int delcount=0; //could be more than one
1295 for (int i=0;i<exon.attrs->Count();i++) {
1296 if (aid==exon.attrs->Get(i)->attr_id) {
1297 if (attrval==NULL ||
1298 strcmp(attrval, exon.attrs->Get(i)->attr_val)==0) {
1299 delcount++;
1300 exon.attrs->freeItem(i);
1301 }
1302 }
1303 }
1304 if (delcount>0) exon.attrs->Pack();
1305 return delcount;
1306 }
1307
1308
1309 void GffObj::getCDS_ends(uint& cds_start, uint& cds_end) {
1310 cds_start=0;
1311 cds_end=0;
1312 if (CDstart==0 || CDend==0) return; //no CDS info
1313 int cdsadj=0;
1314 if (CDphase=='1' || CDphase=='2') {
1315 cdsadj=CDphase-'0';
1316 }
1317 cds_start=CDstart;
1318 cds_end=CDend;
1319 if (strand=='-') cds_end-=cdsadj;
1320 else cds_start+=cdsadj;
1321 }
1322
1323 void GffObj::mRNA_CDS_coords(uint& cds_mstart, uint& cds_mend) {
1324 //sets cds_start and cds_end to the CDS start,end coordinates on the spliced mRNA transcript
1325 cds_mstart=0;
1326 cds_mend=0;
1327 if (CDstart==0 || CDend==0) return; //no CDS info
1328 //restore normal coordinates, just in case
1329 unxcoord();
1330 int cdsadj=0;
1331 if (CDphase=='1' || CDphase=='2') {
1332 cdsadj=CDphase-'0';
1333 }
1334 /*
1335 uint seqstart=CDstart;
1336 uint seqend=CDend;
1337 */
1338 uint seqstart=exons.First()->start;
1339 uint seqend=exons.Last()->end;
1340 int s=0; //resulting nucleotide counter
1341 if (strand=='-') {
1342 for (int x=exons.Count()-1;x>=0;x--) {
1343 uint sgstart=exons[x]->start;
1344 uint sgend=exons[x]->end;
1345 if (seqend<sgstart || seqstart>sgend) continue;
1346 if (seqstart>=sgstart && seqstart<=sgend)
1347 sgstart=seqstart; //seqstart within this segment
1348 if (seqend>=sgstart && seqend<=sgend)
1349 sgend=seqend; //seqend within this segment
1350 s+=(int)(sgend-sgstart)+1;
1351 if (CDstart>=sgstart && CDstart<=sgend) {
1352 //CDstart in this segment
1353 //and we are getting the whole transcript
1354 cds_mend=s-(int)(CDstart-sgstart);
1355 }
1356 if (CDend>=sgstart && CDend<=sgend) {
1357 //CDstart in this segment
1358 //and we are getting the whole transcript
1359 cds_mstart=s-(int)(CDend-cdsadj-sgstart);
1360 }
1361 } //for each exon
1362 } // - strand
1363 else { // + strand
1364 for (int x=0;x<exons.Count();x++) {
1365 uint sgstart=exons[x]->start;
1366 uint sgend=exons[x]->end;
1367 if (seqend<sgstart || seqstart>sgend) continue;
1368 if (seqstart>=sgstart && seqstart<=sgend)
1369 sgstart=seqstart; //seqstart within this segment
1370 if (seqend>=sgstart && seqend<=sgend)
1371 sgend=seqend; //seqend within this segment
1372 s+=(int)(sgend-sgstart)+1;
1373 /* for (uint i=sgstart;i<=sgend;i++) {
1374 spliced[s]=gsubseq[i-gstart];
1375 s++;
1376 }//for each nt
1377 */
1378 if (CDstart>=sgstart && CDstart<=sgend) {
1379 //CDstart in this segment
1380 cds_mstart=s-(int)(sgend-CDstart-cdsadj);
1381 }
1382 if (CDend>=sgstart && CDend<=sgend) {
1383 //CDend in this segment
1384 cds_mend=s-(int)(sgend-CDend);
1385 }
1386 } //for each exon
1387 } // + strand
1388 //spliced[s]=0;
1389 //if (rlen!=NULL) *rlen=s;
1390 //return spliced;
1391 }
1392
1393 char* GffObj::getUnspliced(GFaSeqGet* faseq, int* rlen, GList<GSeg>* seglst)
1394 {
1395 if (faseq==NULL) { GMessage("Warning: getUnspliced(NULL,.. ) called!\n");
1396 return NULL;
1397 }
1398 //restore normal coordinates:
1399 unxcoord();
1400 if (exons.Count()==0) return NULL;
1401 int fspan=end-start+1;
1402 const char* gsubseq=faseq->subseq(start, fspan);
1403 if (gsubseq==NULL) {
1404 GError("Error getting subseq for %s (%d..%d)!\n", gffID, start, end);
1405 }
1406 char* unspliced=NULL;
1407
1408 int seqstart=exons.First()->start;
1409 int seqend=exons.Last()->end;
1410
1411 int unsplicedlen = 0;
1412
1413 unsplicedlen += seqend - seqstart + 1;
1414
1415 GMALLOC(unspliced, unsplicedlen+1); //allocate more here
1416 //uint seqstart, seqend;
1417
1418 int s = 0; //resulting nucleotide counter
1419 if (strand=='-')
1420 {
1421 if (seglst!=NULL)
1422 seglst->Add(new GSeg(s+1,s+1+seqend-seqstart));
1423 for (int i=seqend;i>=seqstart;i--)
1424 {
1425 unspliced[s] = ntComplement(gsubseq[i-start]);
1426 s++;
1427 }//for each nt
1428 } // - strand
1429 else
1430 { // + strand
1431 if (seglst!=NULL)
1432 seglst->Add(new GSeg(s+1,s+1+seqend-seqstart));
1433 for (int i=seqstart;i<=seqend;i++)
1434 {
1435 unspliced[s]=gsubseq[i-start];
1436 s++;
1437 }//for each nt
1438 } // + strand
1439 //assert(s <= unsplicedlen);
1440 unspliced[s]=0;
1441 if (rlen!=NULL) *rlen=s;
1442 return unspliced;
1443 }
1444
1445 char* GffObj::getSpliced(GFaSeqGet* faseq, bool CDSonly, int* rlen, uint* cds_start, uint* cds_end,
1446 GList<GSeg>* seglst) {
1447 if (CDSonly && CDstart==0) return NULL;
1448 if (faseq==NULL) { GMessage("Warning: getSpliced(NULL,.. ) called!\n");
1449 return NULL;
1450 }
1451 //restore normal coordinates:
1452 unxcoord();
1453 if (exons.Count()==0) return NULL;
1454 int fspan=end-start+1;
1455 const char* gsubseq=faseq->subseq(start, fspan);
1456 if (gsubseq==NULL) {
1457 GError("Error getting subseq for %s (%d..%d)!\n", gffID, start, end);
1458 }
1459 if (fspan<(int)(end-start+1)) { //special case: stop coordinate was extended past the gseq length, must adjust
1460 int endadj=end-start+1-fspan;
1461 uint prevend=end;
1462 end-=endadj;
1463 if (CDend>end) CDend=end;
1464 if (exons.Last()->end>end) {
1465 exons.Last()->end=end; //this could get us into trouble if exon start is also > end
1466 if (exons.Last()->start>exons.Last()->end) {
1467 GError("GffObj::getSpliced() error: improper genomic coordinate %d on %s for %s\n",
1468 prevend,getGSeqName(), getID());
1469 }
1470 covlen-=endadj;
1471 }
1472 }
1473 char* spliced=NULL;
1474 GMALLOC(spliced, covlen+1); //allocate more here
1475 uint seqstart, seqend;
1476 int cdsadj=0;
1477 if (CDphase=='1' || CDphase=='2') {
1478 cdsadj=CDphase-'0';
1479 }
1480 if (CDSonly) {
1481 seqstart=CDstart;
1482 seqend=CDend;
1483 if (strand=='-') seqend-=cdsadj;
1484 else seqstart+=cdsadj;
1485 }
1486 else {
1487 seqstart=exons.First()->start;
1488 seqend=exons.Last()->end;
1489 }
1490 int s=0; //resulting nucleotide counter
1491 if (strand=='-') {
1492 for (int x=exons.Count()-1;x>=0;x--) {
1493 uint sgstart=exons[x]->start;
1494 uint sgend=exons[x]->end;
1495 if (seqend<sgstart || seqstart>sgend) continue;
1496 if (seqstart>=sgstart && seqstart<=sgend)
1497 sgstart=seqstart; //seqstart within this segment
1498 if (seqend>=sgstart && seqend<=sgend)
1499 sgend=seqend; //seqend within this segment
1500 if (seglst!=NULL)
1501 seglst->Add(new GSeg(s+1,s+1+sgend-sgstart));
1502 for (uint i=sgend;i>=sgstart;i--) {
1503 spliced[s] = ntComplement(gsubseq[i-start]);
1504 s++;
1505 }//for each nt
1506
1507 if (!CDSonly && cds_start!=NULL && CDstart>0) {
1508 if (CDstart>=sgstart && CDstart<=sgend) {
1509 //CDstart in this segment
1510 //and we are getting the whole transcript
1511 *cds_end=s-(CDstart-sgstart);
1512 }
1513 if (CDend>=sgstart && CDend<=sgend) {
1514 //CDstart in this segment
1515 //and we are getting the whole transcript
1516 *cds_start=s-(CDend-cdsadj-sgstart);
1517 }
1518 }//update local CDS coordinates
1519 } //for each exon
1520 } // - strand
1521 else { // + strand
1522 for (int x=0;x<exons.Count();x++) {
1523 uint sgstart=exons[x]->start;
1524 uint sgend=exons[x]->end;
1525 if (seqend<sgstart || seqstart>sgend) continue;
1526 if (seqstart>=sgstart && seqstart<=sgend)
1527 sgstart=seqstart; //seqstart within this segment
1528 if (seqend>=sgstart && seqend<=sgend)
1529 sgend=seqend; //seqend within this segment
1530 if (seglst!=NULL)
1531 seglst->Add(new GSeg(s+1,s+1+sgend-sgstart));
1532 for (uint i=sgstart;i<=sgend;i++) {
1533 spliced[s]=gsubseq[i-start];
1534 s++;
1535 }//for each nt
1536 if (!CDSonly && cds_start!=NULL && CDstart>0) {
1537 if (CDstart>=sgstart && CDstart<=sgend) {
1538 //CDstart in this segment
1539 //and we are getting the whole transcript
1540 *cds_start=s-(sgend-CDstart-cdsadj);
1541 }
1542 if (CDend>=sgstart && CDend<=sgend) {
1543 //CDstart in this segment
1544 //and we are getting the whole transcript
1545 *cds_end=s-(sgend-CDend);
1546 }
1547 }//update local CDS coordinates
1548 } //for each exon
1549 } // + strand
1550 spliced[s]=0;
1551 if (rlen!=NULL) *rlen=s;
1552 return spliced;
1553 }
1554
1555 char* GffObj::getSplicedTr(GFaSeqGet* faseq, bool CDSonly, int* rlen) {
1556 if (CDSonly && CDstart==0) return NULL;
1557 //restore normal coordinates:
1558 unxcoord();
1559 if (exons.Count()==0) return NULL;
1560 int fspan=end-start+1;
1561 const char* gsubseq=faseq->subseq(start, fspan);
1562 if (gsubseq==NULL) {
1563 GError("Error getting subseq for %s (%d..%d)!\n", gffID, start, end);
1564 }
1565
1566 char* translation=NULL;
1567 GMALLOC(translation, (int)(covlen/3)+1);
1568 uint seqstart, seqend;
1569 int cdsadj=0;
1570 if (CDphase=='1' || CDphase=='2') {
1571 cdsadj=CDphase-'0';
1572 }
1573 if (CDSonly) {
1574 seqstart=CDstart;
1575 seqend=CDend;
1576 if (strand=='-') seqend-=cdsadj;
1577 else seqstart+=cdsadj;
1578 }
1579 else {
1580 seqstart=exons.First()->start;
1581 seqend=exons.Last()->end;
1582 }
1583 Codon codon;
1584 int nt=0; //codon nucleotide counter (0..2)
1585 int aa=0; //aminoacid count
1586 if (strand=='-') {
1587 for (int x=exons.Count()-1;x>=0;x--) {
1588 uint sgstart=exons[x]->start;
1589 uint sgend=exons[x]->end;
1590 if (seqend<sgstart || seqstart>sgend) continue;
1591 if (seqstart>=sgstart && seqstart<=sgend)
1592 sgstart=seqstart; //seqstart within this segment
1593 if (seqend>=sgstart && seqend<=sgend) {
1594 sgend=seqend; //seqend within this segment
1595 }
1596 for (uint i=sgend;i>=sgstart;i--) {
1597 codon.nuc[nt]=ntComplement(gsubseq[i-start]);
1598 nt++;
1599 if (nt==3) {
1600 nt=0;
1601 translation[aa]=codon.translate();
1602 aa++;
1603 }
1604 }//for each nt
1605 } //for each exon
1606 } // - strand
1607 else { // + strand
1608 for (int x=0;x<exons.Count();x++) {
1609 uint sgstart=exons[x]->start;
1610 uint sgend=exons[x]->end;
1611 if (seqend<sgstart || seqstart>sgend) continue;
1612 if (seqstart>=sgstart && seqstart<=sgend)
1613 sgstart=seqstart; //seqstart within this segment
1614 if (seqend>=sgstart && seqend<=sgend)
1615 sgend=seqend; //seqend within this segment
1616 for (uint i=sgstart;i<=sgend;i++) {
1617 codon.nuc[nt]=gsubseq[i-start];
1618 nt++;
1619 if (nt==3) {
1620 nt=0;
1621 translation[aa]=codon.translate();
1622 aa++;
1623 }
1624 }//for each nt
1625 } //for each exon
1626 } // + strand
1627 translation[aa]=0;
1628 if (rlen!=NULL) *rlen=aa;
1629 return translation;
1630 }
1631
1632 void GffObj::printSummary(FILE* fout) {
1633 if (fout==NULL) fout=stdout;
1634 fprintf(fout, "%s\t%c\t%d\t%d\t%4.2f\t%4.1f\n", gffID,
1635 strand, start, end, gscore, (float)qcov/10.0);
1636 }
1637
1638 void GffObj::printGxfLine(FILE* fout, const char* tlabel, const char* gseqname, bool iscds,
1639 uint segstart, uint segend, int exidx, char phase, bool gff3) {
1640 static char scorestr[14];
1641 strcpy(scorestr,".");
1642 GffAttrs* xattrs=NULL;
1643 if (exidx>=0) {
1644 if (exons[exidx]->score) sprintf(scorestr,"%.2f", exons[exidx]->score);
1645 xattrs=exons[exidx]->attrs;
1646 }
1647 if (phase==0 || !iscds) phase='.';
1648 const char* ftype=iscds ? "CDS" : getSubfName();
1649 if (gff3) {
1650 fprintf(fout,
1651 "%s\t%s\t%s\t%d\t%d\t%s\t%c\t%c\tParent=%s",
1652 gseqname, tlabel, ftype, segstart, segend, scorestr, strand,
1653 phase, gffID);
1654 if (xattrs!=NULL) {
1655 for (int i=0;i<xattrs->Count();i++)
1656 fprintf(fout, ";%s=%s",names->attrs.getName(xattrs->Get(i)->attr_id),
1657 xattrs->Get(i)->attr_val);
1658 }
1659 fprintf(fout, "\n");
1660 } //GFF
1661 else {//for GTF -- we print only transcripts
1662 //if (isValidTranscript())
1663 fprintf(fout, "%s\t%s\t%s\t%d\t%d\t%s\t%c\t%c\ttranscript_id \"%s\";",
1664 gseqname, tlabel, ftype, segstart, segend, scorestr, strand, phase, gffID);
1665 //char* geneid=(geneID!=NULL)? geneID : gffID;
1666 if (geneID)
1667 fprintf(fout," gene_id \"%s\";",geneID);
1668 if (gene_name!=NULL) {
1669 //fprintf(fout, " gene_name ");
1670 //if (gene_name[0]=='"') fprintf (fout, "%s;",gene_name);
1671 // else fprintf(fout, "\"%s\";",gene_name);
1672 fprintf(fout," gene_name \"%s\";",gene_name);
1673 }
1674 if (xattrs!=NULL) {
1675 for (int i=0;i<xattrs->Count();i++) {
1676 if (xattrs->Get(i)->attr_val==NULL) continue;
1677 const char* attrname=names->attrs.getName(xattrs->Get(i)->attr_id);
1678 fprintf(fout, " %s ",attrname);
1679 if (xattrs->Get(i)->attr_val[0]=='"')
1680 fprintf(fout, "%s;",xattrs->Get(i)->attr_val);
1681 else fprintf(fout, "\"%s\";",xattrs->Get(i)->attr_val);
1682 }
1683 }
1684 //for GTF, also append the GffObj attributes to each exon line
1685 if ((xattrs=this->attrs)!=NULL) {
1686 for (int i=0;i<xattrs->Count();i++) {
1687 if (xattrs->Get(i)->attr_val==NULL) continue;
1688 const char* attrname=names->attrs.getName(xattrs->Get(i)->attr_id);
1689 fprintf(fout, " %s ",attrname);
1690 if (xattrs->Get(i)->attr_val[0]=='"')
1691 fprintf(fout, "%s;",xattrs->Get(i)->attr_val);
1692 else fprintf(fout, "\"%s\";",xattrs->Get(i)->attr_val);
1693 }
1694 }
1695 fprintf(fout, "\n");
1696 }//GTF
1697 }
1698
1699 void GffObj::printGxf(FILE* fout, GffPrintMode gffp,
1700 const char* tlabel, const char* gfparent) {
1701 static char tmpstr[255];
1702 if (tlabel==NULL) {
1703 tlabel=track_id>=0 ? names->tracks.Get(track_id)->name :
1704 (char*)"gffobj" ;
1705 }
1706 unxcoord();
1707 //if (exons.Count()==0) return;
1708 const char* gseqname=names->gseqs.Get(gseq_id)->name;
1709 bool gff3 = (gffp>=pgffAny);
1710 bool showCDS = (gffp==pgtfAny || gffp==pgtfCDS || gffp==pgffCDS || gffp==pgffAny || gffp==pgffBoth);
1711 bool showExon = (gffp<=pgtfExon || gffp==pgffAny || gffp==pgffExon || gffp==pgffBoth);
1712 if (gff3) {
1713 //print GFF3 mRNA line:
1714 if (gscore>0.0) sprintf(tmpstr,"%.2f", gscore);
1715 else strcpy(tmpstr,".");
1716 uint pstart, pend;
1717 if (gffp==pgffCDS) {
1718 pstart=CDstart;
1719 pend=CDend;
1720 }
1721 else { pstart=start;pend=end; }
1722 //const char* ftype=isTranscript() ? "mRNA" : getFeatureName();
1723 const char* ftype=getFeatureName();
1724 fprintf(fout,
1725 "%s\t%s\t%s\t%d\t%d\t%s\t%c\t.\tID=%s",
1726 gseqname, tlabel, ftype, pstart, pend, tmpstr, strand, gffID);
1727 if (CDstart>0 && !showCDS && !isCDS) fprintf(fout,";CDS=%d-%d",CDstart,CDend);
1728 if (gfparent!=NULL) {
1729 //parent override
1730 fprintf(fout, ";Parent=%s",gfparent);
1731 }
1732 else {
1733 if (parent!=NULL && !parent->isDiscarded())
1734 fprintf(fout, ";Parent=%s",parent->getID());
1735 }
1736 if (geneID!=NULL)
1737 fprintf(fout, ";geneID=%s",geneID);
1738 if (gene_name!=NULL)
1739 fprintf(fout, ";gene_name=%s",gene_name);
1740 if (attrs!=NULL) {
1741 for (int i=0;i<attrs->Count();i++) {
1742 const char* attrname=names->attrs.getName(attrs->Get(i)->attr_id);
1743 fprintf(fout,";%s=%s", attrname,
1744 attrs->Get(i)->attr_val);
1745 }
1746 }
1747 fprintf(fout,"\n");
1748 }// gff3 mRNA line
1749 if (showExon) {
1750 //print exons
1751 if (isCDS && exons.Count()>0 &&
1752 ((strand=='-' && exons.Last()->phase<'0') || (strand=='+' && exons.Last()->phase<'0')))
1753 updateExonPhase();
1754
1755 for (int i=0;i<exons.Count();i++) {
1756 printGxfLine(fout, tlabel, gseqname, isCDS, exons[i]->start, exons[i]->end, i, exons[i]->phase, gff3);
1757 }
1758 }//printing exons
1759 if (showCDS && !isCDS && CDstart>0) {
1760 GArray<GffCDSeg> cds(true,true);
1761 getCDSegs(cds);
1762 for (int i=0;i<cds.Count();i++) {
1763 printGxfLine(fout, tlabel, gseqname, true, cds[i].start, cds[i].end, -1, cds[i].phase, gff3);
1764 }
1765 } //showCDS
1766 }
1767
1768 void GffObj::updateExonPhase() {
1769 if (!isCDS) return;
1770 int cdsacc=0;
1771 if (CDphase=='1' || CDphase=='2') {
1772 cdsacc+= 3-(CDphase-'0');
1773 }
1774 if (strand=='-') { //reverse strand
1775 for (int i=exons.Count()-1;i>=0;i--) {
1776 exons[i]->phase='0'+ (3-cdsacc%3)%3;
1777 cdsacc+=exons[i]->end-exons[i]->start+1;
1778 }
1779 }
1780 else { //forward strand
1781 for (int i=0;i<exons.Count();i++) {
1782 exons[i]->phase='0'+ (3-cdsacc%3)%3;
1783 cdsacc+=exons[i]->end-exons[i]->start+1;
1784 }
1785 }
1786 }
1787
1788
1789 void GffObj::getCDSegs(GArray<GffCDSeg>& cds) {
1790 GffCDSeg cdseg;
1791 int cdsacc=0;
1792 if (CDphase=='1' || CDphase=='2') {
1793 cdsacc+= 3-(CDphase-'0');
1794 }
1795 if (strand=='-') {
1796 for (int x=exons.Count()-1;x>=0;x--) {
1797 uint sgstart=exons[x]->start;
1798 uint sgend=exons[x]->end;
1799 if (CDend<sgstart || CDstart>sgend) continue;
1800 if (CDstart>=sgstart && CDstart<=sgend)
1801 sgstart=CDstart; //cdstart within this segment
1802 if (CDend>=sgstart && CDend<=sgend)
1803 sgend=CDend; //cdend within this segment
1804 cdseg.start=sgstart;
1805 cdseg.end=sgend;
1806 cdseg.exonidx=x;
1807 //cdseg.phase='0'+(cdsacc>0 ? (3-cdsacc%3)%3 : 0);
1808 cdseg.phase='0'+ (3-cdsacc%3)%3;
1809 cdsacc+=sgend-sgstart+1;
1810 cds.Add(cdseg);
1811 } //for each exon
1812 } // - strand
1813 else { // + strand
1814 for (int x=0;x<exons.Count();x++) {
1815 uint sgstart=exons[x]->start;
1816 uint sgend=exons[x]->end;
1817 if (CDend<sgstart || CDstart>sgend) continue;
1818 if (CDstart>=sgstart && CDstart<=sgend)
1819 sgstart=CDstart; //seqstart within this segment
1820 if (CDend>=sgstart && CDend<=sgend)
1821 sgend=CDend; //seqend within this segment
1822 cdseg.start=sgstart;
1823 cdseg.end=sgend;
1824 cdseg.exonidx=x;
1825 //cdseg.phase='0'+(cdsacc>0 ? (3-cdsacc%3)%3 : 0);
1826 cdseg.phase='0' + (3-cdsacc%3)%3 ;
1827 cdsacc+=sgend-sgstart+1;
1828 cds.Add(cdseg);
1829 } //for each exon
1830 } // + strand
1831 }