ViewVC Help
View File | Revision Log | Show Annotations | View Changeset | Root Listing
root/gclib/tophat_cpp/common.cpp
Revision: 67
Committed: Mon Sep 19 19:39:49 2011 UTC (9 years, 5 months ago) by gpertea
File size: 29430 byte(s)
Log Message:
fixed segment_juncs to work with .bam files

Line File contents
1 /*
2 * common.cpp
3 * TopHat
4 *
5 * Created by Cole Trapnell on 11/26/08.
6 * Copyright 2008 Cole Trapnell. All rights reserved.
7 *
8 */
9
10 #ifdef HAVE_CONFIG_H
11 #include <config.h>
12 #endif
13
14 #include <iostream>
15 #include <sstream>
16 #include <cstdarg>
17 #include <getopt.h>
18
19 #include "common.h"
20
21 using namespace std;
22
23 #ifdef MEM_DEBUG
24 //function for debugging memory usage of current program in Linux
25
26 #include <unistd.h>
27 #include <ios>
28 #include <fstream>
29
30 //////////////////////////////////////////////////////////////////////////////
31 // process_mem_usage(double &, double &) - takes two doubles by reference,
32 // attempts to read the system-dependent data for a process' virtual memory
33 // size and resident set size, and return the results in KB.
34 //
35 // On failure, returns 0.0, 0.0
36
37 void process_mem_usage(double& vm_usage, double& resident_set) {
38 using std::ios_base;
39 using std::ifstream;
40 using std::string;
41 vm_usage = 0.0;
42 resident_set = 0.0;
43 // 'file' stat seems to give the most reliable results
44 ifstream stat_stream("/proc/self/stat",ios_base::in);
45 // dummy vars for leading entries in stat that we don't care about
46 string pid, comm, state, ppid, pgrp, session, tty_nr;
47 string tpgid, flags, minflt, cminflt, majflt, cmajflt;
48 string utime, stime, cutime, cstime, priority, nice;
49 string O, itrealvalue, starttime;
50
51 // the two fields we want
52 //
53 unsigned long vsize;
54 long rss;
55
56 stat_stream >> pid >> comm >> state >> ppid >> pgrp >> session >> tty_nr
57 >> tpgid >> flags >> minflt >> cminflt >> majflt >> cmajflt
58 >> utime >> stime >> cutime >> cstime >> priority >> nice
59 >> O >> itrealvalue >> starttime >> vsize >> rss; // don't care about the rest
60
61 stat_stream.close();
62
63 long page_size_kb = sysconf(_SC_PAGE_SIZE) / 1024; // in case x86-64 is configured to use 2MB pages
64 vm_usage = vsize / 1024.0;
65 resident_set = rss * page_size_kb;
66 }
67
68 void print_mem_usage() {
69 double vs, rs;
70 process_mem_usage(vs,rs);
71 vs/=1024;
72 rs/=1024;
73 fprintf(stderr, "VMSize: %6.1fMB\tRSize: %6.1fMB\n", vs, rs);
74 }
75 #endif
76
77
78 unsigned int max_insertion_length = 3;
79 unsigned int max_deletion_length = 3;
80
81
82 int inner_dist_mean = 200;
83 int inner_dist_std_dev = 20;
84 int max_mate_inner_dist = -1;
85
86 int min_anchor_len = 8;
87 int min_report_intron_length = 50;
88 int max_report_intron_length = 500000;
89
90 int min_closure_intron_length = 50;
91 int max_closure_intron_length = 5000;
92
93 int min_coverage_intron_length = 50;
94 int max_coverage_intron_length = 20000;
95
96 int min_segment_intron_length = 50;
97 int max_segment_intron_length = 500000;
98
99 uint32_t min_closure_exon_length = 100;
100
101 int island_extension = 25;
102 int segment_length = 25;
103 int segment_mismatches = 2;
104
105 int max_splice_mismatches = 1;
106
107 ReadFormat reads_format = FASTQ;
108
109 bool verbose = false;
110
111 int max_multihits = 40;
112 bool no_closure_search = false;
113 bool no_coverage_search = false;
114 bool no_microexon_search = false;
115 bool butterfly_search = false;
116 int num_cpus = 1;
117 float min_isoform_fraction = 0.15f;
118
119 string output_dir = "tophat_out";
120 string aux_outfile = ""; //auxiliary output file name (e.g. prep_reads read stats)
121 string gene_filter = "";
122 string gff_file = "";
123 string ium_reads = "";
124 string sam_header = "";
125 string sam_readgroup_id = "";
126 string zpacker = "";
127 string samtools_path = "samtools";
128
129 bool solexa_quals = false;
130 bool phred64_quals = false;
131 bool quals = false;
132 bool integer_quals = false;
133 bool color = false;
134 bool color_out = false;
135
136 string gtf_juncs = "";
137
138 eLIBRARY_TYPE library_type = LIBRARY_TYPE_NONE;
139
140 extern void print_usage();
141
142 /**
143 * Parse an int out of optarg and enforce that it be at least 'lower';
144 * if it is less than 'lower', than output the given error message and
145 * exit with an error and a usage message.
146 */
147
148 int parseIntOpt(int lower, const char *errmsg, void (*print_usage)()) {
149 long l;
150 char *endPtr= NULL;
151 l = strtol(optarg, &endPtr, 10);
152 if (endPtr != NULL) {
153 if (l < lower) {
154 cerr << errmsg << endl;
155 print_usage();
156 exit(1);
157 }
158 return (int32_t)l;
159 }
160 cerr << errmsg << endl;
161 print_usage();
162 exit(1);
163 return -1;
164 }
165
166 /**
167 * Parse an int out of optarg and enforce that it be at least 'lower';
168 * if it is less than 'lower', than output the given error message and
169 * exit with an error and a usage message.
170 */
171 static float parseFloatOpt(float lower, float upper, const char *errmsg, void (*print_usage)()) {
172 float l;
173 l = (float)atof(optarg);
174
175 if (l < lower) {
176 cerr << errmsg << endl;
177 print_usage();
178 exit(1);
179 }
180
181 if (l > upper)
182 {
183 cerr << errmsg << endl;
184 print_usage();
185 exit(1);
186 }
187
188 return l;
189
190 cerr << errmsg << endl;
191 print_usage();
192 exit(1);
193 return -1;
194 }
195
196 /*
197 this is from
198 http://www.winehq.org/pipermail/wine-patches/2001-November/001322.html
199 */
200 char* get_token(char** str, const char* delims)
201 {
202 char* token;
203 if (*str == NULL)
204 return NULL;
205
206 token = *str;
207 while (**str != '\0')
208 {
209 if (strchr(delims, **str) != NULL)
210 {
211 **str = '\0';
212 ++(*str);
213 return token;
214 }
215
216 ++(*str);
217 }
218
219 *str = NULL;
220 return token;
221 }
222
223
224 const char *short_options = "QCp:z:";
225
226 enum
227 {
228 OPT_FASTA = 127,
229 OPT_FASTQ,
230 OPT_MIN_ANCHOR,
231 OPT_SPLICE_MISMATCHES,
232 OPT_VERBOSE,
233 OPT_INSERT_LENGTH_MEAN,
234 OPT_INSERT_LENGTH_STD_DEV,
235 OPT_MIN_ISOFORM_FRACTION,
236 OPT_OUTPUT_DIR,
237 OPT_GENE_FILTER,
238 OPT_GFF_ANNOTATIONS,
239 OPT_MAX_MULTIHITS,
240 OPT_NO_CLOSURE_SEARCH,
241 OPT_NO_COVERAGE_SEARCH,
242 OPT_NO_MICROEXON_SEARCH,
243 OPT_SEGMENT_LENGTH,
244 OPT_SEGMENT_MISMATCHES,
245 OPT_MIN_CLOSURE_EXON,
246 OPT_MAX_CLOSURE_INTRON,
247 OPT_MIN_CLOSURE_INTRON,
248 OPT_MAX_COVERAGE_INTRON,
249 OPT_MIN_COVERAGE_INTRON,
250 OPT_MIN_SEGMENT_INTRON,
251 OPT_MAX_SEGMENT_INTRON,
252 OPT_MIN_REPORT_INTRON,
253 OPT_MAX_REPORT_INTRON,
254 OPT_IUM_READS,
255 OPT_BUTTERFLY_SEARCH,
256 OPT_SOLEXA_QUALS,
257 OPT_PHRED64_QUALS,
258 OPT_SAM_HEADER,
259 OPT_SAM_READGROUP_ID,
260 OPT_QUALS,
261 OPT_INTEGER_QUALS,
262 OPT_COLOR,
263 OPT_COLOR_OUT,
264 OPT_LIBRARY_TYPE,
265 OPT_MAX_DELETION_LENGTH,
266 OPT_MAX_INSERTION_LENGTH,
267 OPT_NUM_CPUS,
268 OPT_ZPACKER,
269 OPT_SAMTOOLS,
270 OPT_AUX_OUT,
271 OPT_GTF_JUNCS
272 };
273
274 static struct option long_options[] = {
275 {"fasta", no_argument, 0, OPT_FASTA},
276 {"fastq", no_argument, 0, OPT_FASTQ},
277 {"min-anchor", required_argument, 0, OPT_MIN_ANCHOR},
278 {"sam-header", required_argument, 0, OPT_SAM_HEADER},
279 {"rg-id", required_argument, 0, OPT_SAM_READGROUP_ID},
280 {"splice-mismatches", required_argument, 0, OPT_SPLICE_MISMATCHES},
281 {"verbose", no_argument, 0, OPT_VERBOSE},
282 {"inner-dist-mean", required_argument, 0, OPT_INSERT_LENGTH_MEAN},
283 {"inner-dist-std-dev", required_argument, 0, OPT_INSERT_LENGTH_STD_DEV},
284 {"output-dir", required_argument, 0, OPT_OUTPUT_DIR},
285 {"gene-filter", required_argument, 0, OPT_GENE_FILTER},
286 {"gtf-annotations", required_argument, 0, OPT_GFF_ANNOTATIONS},
287 {"max-multihits", required_argument, 0, OPT_MAX_MULTIHITS},
288 {"no-closure-search", no_argument, 0, OPT_NO_CLOSURE_SEARCH},
289 {"no-coverage-search", no_argument, 0, OPT_NO_COVERAGE_SEARCH},
290 {"no-microexon-search", no_argument, 0, OPT_NO_MICROEXON_SEARCH},
291 {"segment-length", required_argument, 0, OPT_SEGMENT_LENGTH},
292 {"segment-mismatches", required_argument, 0, OPT_SEGMENT_MISMATCHES},
293 {"min-closure-exon", required_argument, 0, OPT_MIN_CLOSURE_EXON},
294 {"min-closure-intron", required_argument, 0, OPT_MIN_CLOSURE_INTRON},
295 {"max-closure-intron", required_argument, 0, OPT_MAX_CLOSURE_INTRON},
296 {"min-coverage-intron", required_argument, 0, OPT_MIN_COVERAGE_INTRON},
297 {"max-coverage-intron", required_argument, 0, OPT_MAX_COVERAGE_INTRON},
298 {"min-segment-intron", required_argument, 0, OPT_MIN_SEGMENT_INTRON},
299 {"max-segment-intron", required_argument, 0, OPT_MAX_SEGMENT_INTRON},
300 {"min-report-intron", required_argument, 0, OPT_MIN_REPORT_INTRON},
301 {"max-report-intron", required_argument, 0, OPT_MAX_REPORT_INTRON},
302 {"min-isoform-fraction",required_argument, 0, OPT_MIN_ISOFORM_FRACTION},
303 {"ium-reads", required_argument, 0, OPT_IUM_READS},
304 {"butterfly-search", no_argument, 0, OPT_BUTTERFLY_SEARCH},
305 {"solexa-quals", no_argument, 0, OPT_SOLEXA_QUALS},
306 {"phred64-quals", no_argument, 0, OPT_PHRED64_QUALS},
307 {"quals", no_argument, 0, OPT_QUALS},
308 {"integer-quals", no_argument, 0, OPT_INTEGER_QUALS},
309 {"color", no_argument, 0, OPT_COLOR},
310 {"color-out", no_argument, 0, OPT_COLOR_OUT},
311 {"library-type", required_argument, 0, OPT_LIBRARY_TYPE},
312 {"max-deletion-length", required_argument, 0, OPT_MAX_DELETION_LENGTH},
313 {"max-insertion-length", required_argument, 0, OPT_MAX_INSERTION_LENGTH},
314 {"num-threads", required_argument, 0, OPT_NUM_CPUS},
315 {"zpacker", required_argument, 0, OPT_ZPACKER},
316 {"samtools", required_argument, 0, OPT_SAMTOOLS},
317 {"aux-outfile", required_argument, 0, OPT_AUX_OUT},
318 {"gtf-juncs", required_argument, 0, OPT_GTF_JUNCS},
319 {0, 0, 0, 0} // terminator
320 };
321
322
323 void str_appendInt(string& str, int v) {
324 stringstream ss;
325 ss << v;
326 str.append(ss.str());
327 }
328
329 bool str_endsWith(string& str, const char* suffix) {
330 if (str.empty() || str.length()<3) return false;
331 size_t l=strlen(suffix);
332 if (str.length()<=l) return false;
333 if (str.rfind(suffix, str.length()-l-1)!=string::npos) return true;
334 return false;
335 }
336
337 int parse_options(int argc, char** argv, void (*print_usage)())
338 {
339 int option_index = 0;
340 int next_option;
341 do {
342 next_option = getopt_long(argc, argv, short_options, long_options, &option_index);
343 switch (next_option) {
344 case -1:
345 break;
346 case OPT_FASTA:
347 reads_format = FASTA;
348 break;
349 case OPT_FASTQ:
350 reads_format = FASTQ;
351 break;
352 case OPT_MIN_ANCHOR:
353 min_anchor_len = (uint32_t)parseIntOpt(3, "--min-anchor arg must be at least 3", print_usage);
354 break;
355 case OPT_SPLICE_MISMATCHES:
356 max_splice_mismatches = parseIntOpt(0, "--splice-mismatches arg must be at least 0", print_usage);
357 break;
358 case OPT_VERBOSE:
359 verbose = true;
360 break;
361 case OPT_INSERT_LENGTH_MEAN:
362 inner_dist_mean = parseIntOpt(-1024, "--inner-dist-mean arg must be at least -1024", print_usage);
363 break;
364 case OPT_INSERT_LENGTH_STD_DEV:
365 inner_dist_std_dev = parseIntOpt(0, "--inner-dist-std-dev arg must be at least 0", print_usage);
366 break;
367 case OPT_OUTPUT_DIR:
368 output_dir = optarg;
369 break;
370 case OPT_GENE_FILTER:
371 gene_filter = optarg;
372 break;
373 case OPT_GFF_ANNOTATIONS:
374 gff_file = optarg;
375 break;
376 case OPT_MAX_MULTIHITS:
377 max_multihits = parseIntOpt(1, "--max-multihits arg must be at least 1", print_usage);
378 break;
379 case OPT_NO_CLOSURE_SEARCH:
380 no_closure_search = true;
381 break;
382 case OPT_NO_COVERAGE_SEARCH:
383 no_coverage_search = true;
384 break;
385 case OPT_NO_MICROEXON_SEARCH:
386 no_microexon_search = true;
387 break;
388 case OPT_SEGMENT_LENGTH:
389 segment_length = parseIntOpt(4, "--segment-length arg must be at least 4", print_usage);
390 break;
391 case OPT_SEGMENT_MISMATCHES:
392 segment_mismatches = parseIntOpt(0, "--segment-mismatches arg must be at least 0", print_usage);
393 break;
394 case OPT_MIN_CLOSURE_EXON:
395 min_closure_exon_length = parseIntOpt(1, "--min-closure-exon arg must be at least 1", print_usage);
396 break;
397 case OPT_MIN_CLOSURE_INTRON:
398 min_closure_intron_length = parseIntOpt(1, "--min-closure-intron arg must be at least 1", print_usage);
399 break;
400 case OPT_MAX_CLOSURE_INTRON:
401 max_closure_intron_length = parseIntOpt(1, "--max-closure-intron arg must be at least 1", print_usage);
402 break;
403 case OPT_MIN_COVERAGE_INTRON:
404 min_coverage_intron_length = parseIntOpt(1, "--min-coverage-intron arg must be at least 1", print_usage);
405 break;
406 case OPT_MAX_COVERAGE_INTRON:
407 max_coverage_intron_length = parseIntOpt(1, "--max-coverage-intron arg must be at least 1", print_usage);
408 break;
409 case OPT_MIN_SEGMENT_INTRON:
410 min_segment_intron_length = parseIntOpt(1, "--min-segment-intron arg must be at least 1", print_usage);
411 break;
412 case OPT_MAX_SEGMENT_INTRON:
413 max_segment_intron_length = parseIntOpt(1, "--max-segment-intron arg must be at least 1", print_usage);
414 break;
415 case OPT_MIN_REPORT_INTRON:
416 min_report_intron_length = parseIntOpt(1, "--min-report-intron arg must be at least 1", print_usage);
417 break;
418 case OPT_MAX_REPORT_INTRON:
419 max_report_intron_length = parseIntOpt(1, "--max-report-intron arg must be at least 1", print_usage);
420 break;
421 case OPT_MIN_ISOFORM_FRACTION:
422 min_isoform_fraction = parseFloatOpt(0.0f, 1.0f, "--min-isoform-fraction arg must be [0.0,1.0]", print_usage);
423 break;
424 case OPT_IUM_READS:
425 ium_reads = optarg;
426 break;
427 case OPT_SAM_HEADER:
428 sam_header = optarg;
429 break;
430 case OPT_SAM_READGROUP_ID:
431 sam_readgroup_id = optarg;
432 break;
433 case OPT_BUTTERFLY_SEARCH:
434 butterfly_search = true;
435 break;
436 case OPT_SOLEXA_QUALS:
437 solexa_quals = true;
438 break;
439 case OPT_PHRED64_QUALS:
440 phred64_quals = true;
441 break;
442 case 'Q':
443 case OPT_QUALS:
444 quals = true;
445 break;
446 case OPT_INTEGER_QUALS:
447 integer_quals = true;
448 break;
449 case 'C':
450 case OPT_COLOR:
451 color = true;
452 break;
453 case OPT_COLOR_OUT:
454 color_out = true;
455 break;
456 case OPT_LIBRARY_TYPE:
457 if (strcmp(optarg, "fr-unstranded") == 0)
458 library_type = FR_UNSTRANDED;
459 else if (strcmp(optarg, "fr-firststrand") == 0)
460 library_type = FR_FIRSTSTRAND;
461 else if (strcmp(optarg, "fr-secondstrand") == 0)
462 library_type = FR_SECONDSTRAND;
463 else if (strcmp(optarg, "ff-unstranded") == 0)
464 library_type = FF_UNSTRANDED;
465 else if (strcmp(optarg, "ff-firststrand") == 0)
466 library_type = FF_FIRSTSTRAND;
467 else if (strcmp(optarg, "ff-secondstrand") == 0)
468 library_type = FF_SECONDSTRAND;
469 break;
470 case OPT_MAX_DELETION_LENGTH:
471 max_deletion_length = parseIntOpt(0, "--max-deletion-length must be at least 0", print_usage);
472 break;
473 case OPT_MAX_INSERTION_LENGTH:
474 max_insertion_length = parseIntOpt(0, "--max-insertion-length must be at least 0", print_usage);
475 break;
476 case 'z':
477 case OPT_ZPACKER:
478 zpacker = optarg;
479 break;
480 case OPT_SAMTOOLS:
481 samtools_path = optarg;
482 break;
483 case OPT_AUX_OUT:
484 aux_outfile = optarg;
485 break;
486 case 'p':
487 case OPT_NUM_CPUS:
488 num_cpus=parseIntOpt(1,"-p/--num-threads must be at least 1",print_usage);
489 break;
490 case OPT_GTF_JUNCS:
491 gtf_juncs = optarg;
492 break;
493 default:
494 print_usage();
495 return 1;
496 }
497 } while(next_option != -1);
498
499 return 0;
500 }
501
502
503 // Error routine (prints error message and exits!)
504 void err_exit(const char* format,...){
505 va_list arguments;
506 va_start(arguments,format);
507 vfprintf(stderr,format,arguments);
508 va_end(arguments);
509 #ifdef DEBUG
510 // trigger a core dump for later inspection
511 abort();
512 #endif
513 exit(1);
514 }
515
516 FILE* FZPipe::openRead(const char* fname, string& popencmd) {
517 pipecmd=popencmd;
518 filename=fname;
519 if (pipecmd.empty()) {
520 file=fopen(filename.c_str(), "r");
521 }
522 else {
523 string pcmd(pipecmd);
524 pcmd.append(" '");
525 pcmd.append(filename);
526 pcmd.append("'");
527 file=popen(pcmd.c_str(), "r");
528 }
529 return file;
530 }
531
532 FILE* FZPipe::openRead(const char* fname) {
533 string pcmd;
534 return this->openRead(fname,pcmd);
535 }
536
537 FILE* FZPipe::openWrite(const char* fname, string& popencmd) {
538 pipecmd=popencmd;
539 filename=fname;
540 if (pipecmd.empty()) {
541 file=fopen(filename.c_str(), "w");
542 }
543 else {
544 string pcmd(pipecmd);
545 pcmd.append(" - > '");
546 pcmd.append(filename.c_str());
547 pcmd.append("'");
548 file=popen(pcmd.c_str(), "w");
549 }
550 return file;
551 }
552
553
554 FILE* FZPipe::openWrite(const char* fname) {
555 string pcmd;
556 return this->openWrite(fname,pcmd);
557 }
558
559 void FZPipe::rewind() {
560 if (is_bam) return; //does nothing for BAM mappings
561 if (pipecmd.empty()) {
562 if (file!=NULL) {
563 ::rewind(file);
564 return;
565 }
566 if (!filename.empty()) {
567 file=fopen(filename.c_str(),"r");
568 return;
569 }
570 }
571 if (filename.empty())
572 err_die("Error: FZStream::rewind() failed (missing filename)!\n");
573 this->close();
574 string pcmd(pipecmd);
575 pcmd.append(" '");
576 pcmd.append(filename);
577 pcmd.append("'");
578 file=popen(pcmd.c_str(), "r");
579 if (file==NULL) {
580 err_die("Error: FZStream::rewind() popen(%s) failed!\n",pcmd.c_str());
581 }
582 }
583
584
585 string getFext(const string& s) {
586 string r("");
587 //if (xpos!=NULL) *xpos=0;
588 if (s.empty() || s=="-") return r;
589 int slen=(int)s.length();
590 int p=s.rfind('.');
591 int d=s.rfind('/');
592 if (p<=0 || p>slen-2 || p<slen-7 || p<d) return r;
593 r=s.substr(p+1);
594 //if (xpos!=NULL) *xpos=p+1;
595 for(size_t i=0; i!=r.length(); i++)
596 r[i] = std::tolower(r[i]);
597 return r;
598 }
599
600 string guess_packer(const string& fname, bool use_all_cpus) {
601 //only needed for the primary input files (given by user)
602 string picmd("");
603 string fext=getFext(fname);
604 if (fext=="bam") {
605 picmd="bam2fastx";
606 return picmd;
607 }
608 if (fext=="gz" || fext=="gzip" || fext=="z") {
609 if (use_all_cpus && str_endsWith(zpacker,"pigz")) {
610 picmd=zpacker;
611 if (num_cpus<2) picmd.append(" -p1");
612 else {
613 picmd.append(" -p");
614 str_appendInt(picmd, num_cpus);
615 //picmd.append(" -cd");
616 }
617 }
618 else picmd="gzip";
619 }
620 else if (fext=="bz2" || fext=="bzip2" || fext=="bz" || fext=="bzip") {
621 if (use_all_cpus && str_endsWith(zpacker,"pbzip2")) {
622 picmd=zpacker;
623 if (num_cpus<2) picmd.append(" -p1");
624 else {
625 picmd.append(" -p");
626 str_appendInt(picmd, num_cpus);
627 //picmd.append(" -cd");
628 }
629 }
630 else picmd="bzip2";
631 }
632 return picmd;
633 }
634
635 /*
636 string getBam2SamCmd(const string& fname) {
637 string pipecmd("");
638 string fext=getFext(fname);
639 if (fext=="bam") {
640 pipecmd=samtools_path;
641 pipecmd.append(" view");
642 }
643 return pipecmd;
644 }
645 */
646
647 void err_die(const char* format,...) { // Error exit
648 va_list arguments;
649 va_start(arguments,format);
650 vfprintf(stderr,format,arguments);
651 va_end(arguments);
652 exit(1);
653 }
654
655 string getUnpackCmd(const string& fname, bool use_all_cpus) {
656 //prep_reads should use guess_packer() instead
657 string pipecmd("");
658 string fext=getFext(fname);
659 if (fext=="bam") {
660 pipecmd="bam2fastx";
661 return pipecmd;
662 }
663 if (zpacker.empty() || fext!="z") {
664 return pipecmd;
665 }
666 pipecmd=zpacker;
667 if (str_endsWith(pipecmd, "pigz") ||str_endsWith(pipecmd, "pbzip2")) {
668 if (use_all_cpus==false) pipecmd.append(" -p1");
669 else if (num_cpus>1) {
670 pipecmd.append(" -p");
671 str_appendInt(pipecmd,num_cpus);
672 }
673 }
674 if (!pipecmd.empty()) pipecmd.append(" -cd");
675 return pipecmd;
676 }
677
678 void checkSamHeader() {
679 if (sam_header.empty())
680 err_die("Error: writeSamHeader() with empty sam_header string\n");
681 //copy the SAM header
682 FILE* fh=fopen(sam_header.c_str(), "r");
683 if (fh==NULL)
684 err_die("Error: cannot open SAM header file %s\n",sam_header.c_str());
685 fclose(fh);
686 }
687
688 void writeSamHeader(FILE* fout) {
689 if (fout==NULL)
690 err_die("Error: writeSamHeader(NULL)\n");
691 checkSamHeader();
692 //copy the SAM header
693 FILE* fh=fopen(sam_header.c_str(), "r");
694 int ch=-1;
695 while ((ch=fgetc(fh))!=EOF) {
696 if (fputc(ch, fout)==EOF)
697 err_die("Error copying SAM header\n");
698 }
699 fclose(fh);
700 }
701
702 //auxiliary functions for BAM record handling
703 uint8_t* realloc_bdata(bam1_t *b, int size) {
704 if (b->m_data < size) {
705 b->m_data = size;
706 kroundup32(b->m_data);
707 b->data = (uint8_t*)realloc(b->data, b->m_data);
708 }
709 if (b->data_len<size) b->data_len=size;
710 return b->data;
711 }
712
713 uint8_t* dupalloc_bdata(bam1_t *b, int size) {
714 //same as realloc_bdata, but does not free previous data
715 //but returns it instead
716 //it ALWAYS duplicates data
717 b->m_data = size;
718 kroundup32(b->m_data);
719 uint8_t* odata=b->data;
720 b->data = (uint8_t*)malloc(b->m_data);
721 memcpy((void*)b->data, (void*)odata, b->data_len);
722 b->data_len=size;
723 return odata; //user must FREE this after
724 }
725
726 extern unsigned short bam_char2flag_table[];
727
728 GBamRecord::GBamRecord(const char* qname, int32_t gseq_tid,
729 int pos, bool reverse, const char* qseq, const char* cigar, const char* quals) {
730 novel=true;
731 b=bam_init1();
732 b->core.tid=gseq_tid;
733 if (pos<=0) {
734 b->core.pos=-1; //unmapped
735 //if (gseq_tid<0)
736 b->core.flag |= BAM_FUNMAP;
737 }
738 else b->core.pos=pos-1; //BAM is 0-based
739 b->core.qual=255;
740 int l_qseq=strlen(qseq);
741 //this may not be accurate, setting CIGAR is the correct way
742 //b->core.bin = bam_reg2bin(b->core.pos, b->core.pos+l_qseq-1);
743 b->core.l_qname=strlen(qname)+1; //includes the \0 at the end
744 memcpy(realloc_bdata(b, b->core.l_qname), qname, b->core.l_qname);
745 set_cigar(cigar); //this will also set core.bin
746 add_sequence(qseq, l_qseq);
747 add_quals(quals); //quals must be given as Phred33
748 if (reverse) { b->core.flag |= BAM_FREVERSE ; }
749 }
750
751 GBamRecord::GBamRecord(const char* qname, int32_t flags, int32_t g_tid,
752 int pos, int map_qual, const char* cigar, int32_t mg_tid, int mate_pos,
753 int insert_size, const char* qseq, const char* quals,
754 const vector<string>* aux_strings) {
755 novel=true;
756 b=bam_init1();
757 b->core.tid=g_tid;
758 b->core.pos = (pos<=0) ? -1 : pos-1; //BAM is 0-based
759 b->core.qual=map_qual;
760 int l_qseq=strlen(qseq);
761 b->core.l_qname=strlen(qname)+1; //includes the \0 at the end
762 memcpy(realloc_bdata(b, b->core.l_qname), qname, b->core.l_qname);
763 set_cigar(cigar); //this will also set core.bin
764 add_sequence(qseq, l_qseq);
765 add_quals(quals); //quals must be given as Phred33
766 set_flags(flags);
767 set_mdata(mg_tid, (int32_t)(mate_pos-1), (int32_t)insert_size);
768 if (aux_strings!=NULL) {
769 for (vector<string>::const_iterator itr=aux_strings->begin();
770 itr!=aux_strings->end(); ++itr) {
771 add_aux(itr->c_str());
772 }
773 }
774 }
775 void GBamRecord::set_cigar(const char* cigar) {
776 //requires b->core.pos and b->core.flag to have been set properly PRIOR to this call
777 int doff=b->core.l_qname;
778 uint8_t* after_cigar=NULL;
779 int after_cigar_len=0;
780 uint8_t* prev_bdata=NULL;
781 if (b->data_len>doff) {
782 //cigar string already allocated, replace it
783 int d=b->core.l_qname + b->core.n_cigar * 4;//offset of after-cigar data
784 after_cigar=b->data+d;
785 after_cigar_len=b->data_len-d;
786 }
787 const char *s;
788 char *t;
789 int i, op;
790 long x;
791 b->core.n_cigar = 0;
792 if (cigar != NULL && strcmp(cigar, "*") != 0) {
793 for (s = cigar; *s; ++s) {
794 if (isalpha(*s)) b->core.n_cigar++;
795 else if (!isdigit(*s)) {
796 err_die("Error: invalid CIGAR character (%s)\n",cigar);
797 }
798 }
799 if (after_cigar_len>0) { //replace/insert into existing full data
800 prev_bdata=dupalloc_bdata(b, doff + b->core.n_cigar * 4 + after_cigar_len);
801 memcpy((void*)(b->data+doff+b->core.n_cigar*4),(void*)after_cigar, after_cigar_len);
802 free(prev_bdata);
803 }
804 else {
805 realloc_bdata(b, doff + b->core.n_cigar * 4);
806 }
807 for (i = 0, s = cigar; i != b->core.n_cigar; ++i) {
808 x = strtol(s, &t, 10);
809 op = toupper(*t);
810 if (op == 'M' || op == '=' || op == 'X') op = BAM_CMATCH;
811 else if (op == 'I') op = BAM_CINS;
812 else if (op == 'D') op = BAM_CDEL;
813 else if (op == 'N') op = BAM_CREF_SKIP;
814 else if (op == 'S') op = BAM_CSOFT_CLIP;
815 else if (op == 'H') op = BAM_CHARD_CLIP;
816 else if (op == 'P') op = BAM_CPAD;
817 else err_die("Error: invalid CIGAR operation (%s)\n",cigar);
818 s = t + 1;
819 bam1_cigar(b)[i] = x << BAM_CIGAR_SHIFT | op;
820 }
821 if (*s) err_die("Error: unmatched CIGAR operation (%s)\n",cigar);
822 b->core.bin = bam_reg2bin(b->core.pos, bam_calend(&b->core, bam1_cigar(b)));
823 } else {//no CIGAR string given
824 if (!(b->core.flag&BAM_FUNMAP)) {
825 fprintf(stderr, "Warning: mapped sequence without CIGAR (%s)\n", (char*)b->data);
826 b->core.flag |= BAM_FUNMAP;
827 }
828 b->core.bin = bam_reg2bin(b->core.pos, b->core.pos + 1);
829 }
830 } //set_cigar()
831
832 void GBamRecord::add_sequence(const char* qseq, int slen) {
833 //must be called AFTER set_cigar (cannot replace existing sequence for now)
834 if (qseq==NULL) return; //should we ever care about this?
835 if (slen<0) slen=strlen(qseq);
836 int doff = b->core.l_qname + b->core.n_cigar * 4;
837 if (strcmp(qseq, "*")!=0) {
838 b->core.l_qseq=slen;
839 if (b->core.n_cigar && b->core.l_qseq != (int32_t)bam_cigar2qlen(&b->core, bam1_cigar(b)))
840 err_die("Error: CIGAR and sequence length are inconsistent!(%s)\n",
841 qseq);
842 uint8_t* p = (uint8_t*)realloc_bdata(b, doff + (b->core.l_qseq+1)/2 + b->core.l_qseq) + doff;
843 //also allocated quals memory
844 memset(p, 0, (b->core.l_qseq+1)/2);
845 for (int i = 0; i < b->core.l_qseq; ++i)
846 p[i/2] |= bam_nt16_table[(int)qseq[i]] << 4*(1-i%2);
847 } else b->core.l_qseq = 0;
848 }
849
850 void GBamRecord::add_quals(const char* quals) {
851 //requires core.l_qseq already set
852 //and must be called AFTER add_sequence(), which also allocates the memory for quals
853 uint8_t* p = b->data+(b->core.l_qname + b->core.n_cigar * 4 + (b->core.l_qseq+1)/2);
854 if (quals==NULL || strcmp(quals, "*") == 0) {
855 for (int i=0;i < b->core.l_qseq; i++) p[i] = 0xff;
856 return;
857 }
858 for (int i=0;i < b->core.l_qseq; i++) p[i] = quals[i]-33;
859 }
860
861 void GBamRecord::add_aux(const char* str) {
862 //requires: being called AFTER add_quals()
863 static char tag[2];
864 static uint8_t abuf[512];
865 //requires: being called AFTER add_quals()
866 int strl=strlen(str);
867 //int doff = b->core.l_qname + b->core.n_cigar*4 + (b->core.l_qseq+1)/2 + b->core.l_qseq + b->l_aux;
868 //int doff0=doff;
869 if (strl < 6 || str[2] != ':' || str[4] != ':')
870 parse_error("missing colon in auxiliary data");
871 tag[0] = str[0]; tag[1] = str[1];
872 uint8_t atype = str[3];
873 uint8_t* adata=abuf;
874 int alen=0;
875 if (atype == 'A' || atype == 'a' || atype == 'c' || atype == 'C') { // c and C for backward compatibility
876 atype='A';
877 alen=1;
878 adata=(uint8_t*)&str[5];
879 }
880 else if (atype == 'I' || atype == 'i') {
881 long long x=(long long)atoll(str + 5);
882 if (x < 0) {
883 if (x >= -127) {
884 atype='c';
885 abuf[0] = (int8_t)x;
886 alen=1;
887 }
888 else if (x >= -32767) {
889 atype = 's';
890 *(int16_t*)abuf = (int16_t)x;
891 alen=2;
892 }
893 else {
894 atype='i';
895 *(int32_t*)abuf = (int32_t)x;
896 alen=4;
897 if (x < -2147483648ll)
898 fprintf(stderr, "Parse warning: integer %lld is out of range.",
899 x);
900 }
901 } else { //x >=0
902 if (x <= 255) {
903 atype = 'C';
904 abuf[0] = (uint8_t)x;
905 alen=1;
906 }
907 else if (x <= 65535) {
908 atype='S';
909 *(uint16_t*)abuf = (uint16_t)x;
910 alen=2;
911 }
912 else {
913 atype='I';
914 *(uint32_t*)abuf = (uint32_t)x;
915 alen=4;
916 if (x > 4294967295ll)
917 fprintf(stderr, "Parse warning: integer %lld is out of range.",
918 x);
919 }
920 }
921 } //integer type
922 else if (atype == 'f') {
923 *(float*)abuf = (float)atof(str + 5);
924 alen = sizeof(float);
925 }
926 else if (atype == 'd') { //?
927 *(float*)abuf = (float)atof(str + 9);
928 alen=8;
929 }
930 else if (atype == 'Z' || atype == 'H') {
931 if (atype == 'H') { // check whether the hex string is valid
932 if ((strl - 5) % 2 == 1) parse_error("length of the hex string not even");
933 for (int i = 0; i < strl - 5; ++i) {
934 int c = toupper(str[5 + i]);
935 if (!((c >= '0' && c <= '9') || (c >= 'A' && c <= 'F')))
936 parse_error("invalid hex character");
937 }
938 }
939 memcpy(abuf, str + 5, strl - 5);
940 abuf[strl-5] = 0;
941 alen=strl-4;
942 } else parse_error("unrecognized aux type");
943 this->add_aux(tag, atype, alen, adata);
944 }//add_aux()