3 Copyright (c) 2008-2010 Genome Research Ltd (GRL).
5 Permission is hereby granted, free of charge, to any person obtaining
6 a copy of this software and associated documentation files (the
7 "Software"), to deal in the Software without restriction, including
8 without limitation the rights to use, copy, modify, merge, publish,
9 distribute, sublicense, and/or sell copies of the Software, and to
10 permit persons to whom the Software is furnished to do so, subject to
11 the following conditions:
13 The above copyright notice and this permission notice shall be
14 included in all copies or substantial portions of the Software.
16 THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
17 EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
18 MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
19 NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
20 BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
21 ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
22 CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
26 /* Contact: Heng Li <lh3@sanger.ac.uk> */
34 BAM library provides I/O and various operations on manipulating files
35 in the BAM (Binary Alignment/Mapping) or SAM (Sequence Alignment/Map)
36 format. It now supports importing from or exporting to TAM, sorting,
37 merging, generating pileup, and quickly retrieval of reads overlapped
38 with a specified region.
40 @copyright Genome Research Ltd.
43 #define BAM_VERSION "0.1.12-r921:124"
51 #define BAM_VIRTUAL_OFFSET16
53 /*! @abstract BAM file handler */
54 typedef BGZF *bamFile;
55 #define bam_open(fn, mode) bgzf_open(fn, mode)
56 #define bam_dopen(fd, mode) bgzf_fdopen(fd, mode)
57 #define bam_close(fp) bgzf_close(fp)
58 #define bam_read(fp, buf, size) bgzf_read(fp, buf, size)
59 #define bam_write(fp, buf, size) bgzf_write(fp, buf, size)
60 #define bam_tell(fp) bgzf_tell(fp)
61 #define bam_seek(fp, pos, dir) bgzf_seek(fp, pos, dir)
63 #define BAM_TRUE_OFFSET
65 typedef gzFile bamFile;
66 #define bam_open(fn, mode) gzopen(fn, mode)
67 #define bam_dopen(fd, mode) gzdopen(fd, mode)
68 #define bam_close(fp) gzclose(fp)
69 #define bam_read(fp, buf, size) gzread(fp, buf, size)
70 /* no bam_write/bam_tell/bam_seek() here */
74 @abstract Structure for the alignment header.
75 @field n_targets number of reference sequences
76 @field target_name names of the reference sequences
77 @field target_len lengths of the referene sequences
78 @field dict header dictionary
79 @field hash hash table for fast name lookup
80 @field rg2lib hash table for @RG-ID -> LB lookup
81 @field l_text length of the plain text in the header
82 @field text plain text
84 @discussion Field hash points to null by default. It is a private
91 void *dict, *hash, *rg2lib;
92 size_t l_text, n_text;
96 /*! @abstract the read is paired in sequencing, no matter whether it is mapped in a pair */
98 /*! @abstract the read is mapped in a proper pair */
99 #define BAM_FPROPER_PAIR 2
100 /*! @abstract the read itself is unmapped; conflictive with BAM_FPROPER_PAIR */
102 /*! @abstract the mate is unmapped */
103 #define BAM_FMUNMAP 8
104 /*! @abstract the read is mapped to the reverse strand */
105 #define BAM_FREVERSE 16
106 /*! @abstract the mate is mapped to the reverse strand */
107 #define BAM_FMREVERSE 32
108 /*! @abstract this is read1 */
109 #define BAM_FREAD1 64
110 /*! @abstract this is read2 */
111 #define BAM_FREAD2 128
112 /*! @abstract not primary alignment */
113 #define BAM_FSECONDARY 256
114 /*! @abstract QC failure */
115 #define BAM_FQCFAIL 512
116 /*! @abstract optical or PCR duplicate */
117 #define BAM_FDUP 1024
123 /*! @abstract defautl mask for pileup */
124 #define BAM_DEF_MASK (BAM_FUNMAP | BAM_FSECONDARY | BAM_FQCFAIL | BAM_FDUP)
126 #define BAM_CORE_SIZE sizeof(bam1_core_t)
129 * Describing how CIGAR operation/length is packed in a 32-bit integer.
131 #define BAM_CIGAR_SHIFT 4
132 #define BAM_CIGAR_MASK ((1 << BAM_CIGAR_SHIFT) - 1)
137 /*! @abstract CIGAR: match */
139 /*! @abstract CIGAR: insertion to the reference */
141 /*! @abstract CIGAR: deletion from the reference */
143 /*! @abstract CIGAR: skip on the reference (e.g. spliced alignment) */
144 #define BAM_CREF_SKIP 3
145 /*! @abstract CIGAR: clip on the read with clipped sequence present in qseq */
146 #define BAM_CSOFT_CLIP 4
147 /*! @abstract CIGAR: clip on the read with clipped sequence trimmed off */
148 #define BAM_CHARD_CLIP 5
149 /*! @abstract CIGAR: padding */
153 @abstract Structure for core alignment information.
154 @field tid chromosome ID, defined by bam_header_t
155 @field pos 0-based leftmost coordinate
156 @field strand strand; 0 for forward and 1 otherwise
157 @field bin bin calculated by bam_reg2bin()
158 @field qual mapping quality
159 @field l_qname length of the query name
160 @field flag bitwise flag
161 @field n_cigar number of CIGAR operations
162 @field l_qseq length of the query sequence (read)
167 uint32_t bin:16, qual:8, l_qname:8;
168 uint32_t flag:16, n_cigar:16;
176 @abstract Structure for one alignment.
177 @field core core information about the alignment
178 @field l_aux length of auxiliary data
179 @field data_len current length of bam1_t::data
180 @field m_data maximum length of bam1_t::data
181 @field data all variable-length data, concatenated; structure: cigar-qname-seq-qual-aux
185 1. qname is zero tailing and core.l_qname includes the tailing '\0'.
186 2. l_qseq is calculated from the total length of an alignment block
187 on reading or from CIGAR.
191 int l_aux, data_len, m_data;
195 typedef struct __bam_iter_t *bam_iter_t;
197 #define bam1_strand(b) (((b)->core.flag&BAM_FREVERSE) != 0)
198 #define bam1_mstrand(b) (((b)->core.flag&BAM_FMREVERSE) != 0)
201 @abstract Get the CIGAR array
202 @param b pointer to an alignment
203 @return pointer to the CIGAR array
205 @discussion In the CIGAR array, each element is a 32-bit integer. The
206 lower 4 bits gives a CIGAR operation and the higher 28 bits keep the
209 #define bam1_cigar(b) ((uint32_t*)((b)->data + (b)->core.l_qname))
212 @abstract Get the name of the query
213 @param b pointer to an alignment
214 @return pointer to the name string, null terminated
216 #define bam1_qname(b) ((char*)((b)->data))
219 @abstract Get query sequence
220 @param b pointer to an alignment
221 @return pointer to sequence
223 @discussion Each base is encoded in 4 bits: 1 for A, 2 for C, 4 for G,
224 8 for T and 15 for N. Two bases are packed in one byte with the base
225 at the higher 4 bits having smaller coordinate on the read. It is
226 recommended to use bam1_seqi() macro to get the base.
228 #define bam1_seq(b) ((b)->data + (b)->core.n_cigar*4 + (b)->core.l_qname)
231 @abstract Get query quality
232 @param b pointer to an alignment
233 @return pointer to quality string
235 #define bam1_qual(b) ((b)->data + (b)->core.n_cigar*4 + (b)->core.l_qname + (((b)->core.l_qseq + 1)>>1))
238 @abstract Get a base on read
239 @param s Query sequence returned by bam1_seq()
240 @param i The i-th position, 0-based
241 @return 4-bit integer representing the base.
243 #define bam1_seqi(s, i) ((s)[(i)/2] >> 4*(1-(i)%2) & 0xf)
246 @abstract Get query sequence and quality
247 @param b pointer to an alignment
248 @return pointer to the concatenated auxiliary data
250 #define bam1_aux(b) ((b)->data + (b)->core.n_cigar*4 + (b)->core.l_qname + (b)->core.l_qseq + ((b)->core.l_qseq + 1)/2)
254 @abstract Round an integer to the next closest power-2 integer.
255 @param x integer to be rounded (in place)
256 @discussion x will be modified.
258 #define kroundup32(x) (--(x), (x)|=(x)>>1, (x)|=(x)>>2, (x)|=(x)>>4, (x)|=(x)>>8, (x)|=(x)>>16, ++(x))
262 @abstract Whether the machine is big-endian; modified only in
265 extern int bam_is_be;
267 /*! @abstract Table for converting a nucleotide character to the 4-bit encoding. */
268 extern unsigned char bam_nt16_table[256];
270 /*! @abstract Table for converting a 4-bit encoded nucleotide to a letter. */
271 extern char *bam_nt16_rev_table;
273 extern char bam_nt16_nt4_table[];
279 /*********************
280 * Low-level SAM I/O *
281 *********************/
283 /*! @abstract TAM file handler */
284 typedef struct __tamFile_t *tamFile;
287 @abstract Open a SAM file for reading, either uncompressed or compressed by gzip/zlib.
288 @param fn SAM file name
289 @return SAM file handler
291 tamFile sam_open(const char *fn);
294 @abstract Close a SAM file handler
295 @param fp SAM file handler
297 void sam_close(tamFile fp);
300 @abstract Read one alignment from a SAM file handler
301 @param fp SAM file handler
302 @param header header information (ordered names of chromosomes)
303 @param b read alignment; all members in b will be updated
304 @return 0 if successful; otherwise negative
306 int sam_read1(tamFile fp, bam_header_t *header, bam1_t *b);
309 @abstract Read header information from a TAB-delimited list file.
310 @param fn_list file name for the list
311 @return a pointer to the header structure
313 @discussion Each line in this file consists of chromosome name and
314 the length of chromosome.
316 bam_header_t *sam_header_read2(const char *fn_list);
319 @abstract Read header from a SAM file (if present)
320 @param fp SAM file handler
321 @return pointer to header struct; 0 if no @SQ lines available
323 bam_header_t *sam_header_read(tamFile fp);
326 @abstract Parse @SQ lines a update a header struct
327 @param h pointer to the header struct to be updated
328 @return number of target sequences
330 @discussion bam_header_t::{n_targets,target_len,target_name} will
331 be destroyed in the first place.
333 int sam_header_parse(bam_header_t *h);
334 int32_t bam_get_tid(const bam_header_t *header, const char *seq_name);
337 @abstract Parse @RG lines a update a header struct
338 @param h pointer to the header struct to be updated
339 @return number of @RG lines
341 @discussion bam_header_t::rg2lib will be destroyed in the first
344 int sam_header_parse_rg(bam_header_t *h);
346 #define sam_write1(header, b) bam_view1(header, b)
349 /********************************
350 * APIs for string dictionaries *
351 ********************************/
353 int bam_strmap_put(void *strmap, const char *rg, const char *lib);
354 const char *bam_strmap_get(const void *strmap, const char *rg);
355 void *bam_strmap_dup(const void*);
356 void *bam_strmap_init();
357 void bam_strmap_destroy(void *strmap);
360 /*********************
361 * Low-level BAM I/O *
362 *********************/
365 @abstract Initialize a header structure.
366 @return the pointer to the header structure
368 @discussion This function also modifies the global variable
371 bam_header_t *bam_header_init();
374 @abstract Destroy a header structure.
375 @param header pointer to the header
377 void bam_header_destroy(bam_header_t *header);
380 @abstract Read a header structure from BAM.
381 @param fp BAM file handler, opened by bam_open()
382 @return pointer to the header structure
384 @discussion The file position indicator must be placed at the
385 beginning of the file. Upon success, the position indicator will
386 be set at the start of the first alignment.
388 bam_header_t *bam_header_read(bamFile fp);
391 @abstract Write a header structure to BAM.
392 @param fp BAM file handler
393 @param header pointer to the header structure
394 @return always 0 currently
396 int bam_header_write(bamFile fp, const bam_header_t *header);
399 @abstract Read an alignment from BAM.
400 @param fp BAM file handler
401 @param b read alignment; all members are updated.
402 @return number of bytes read from the file
404 @discussion The file position indicator must be
405 placed right before an alignment. Upon success, this function
406 will set the position indicator to the start of the next
407 alignment. This function is not affected by the machine
410 int bam_read1(bamFile fp, bam1_t *b);
413 @abstract Write an alignment to BAM.
414 @param fp BAM file handler
415 @param c pointer to the bam1_core_t structure
416 @param data_len total length of variable size data related to
418 @param data pointer to the concatenated data
419 @return number of bytes written to the file
421 @discussion This function is not affected by the machine
424 int bam_write1_core(bamFile fp, const bam1_core_t *c, int data_len, uint8_t *data);
427 @abstract Write an alignment to BAM.
428 @param fp BAM file handler
429 @param b alignment to write
430 @return number of bytes written to the file
432 @abstract It is equivalent to:
433 bam_write1_core(fp, &b->core, b->data_len, b->data)
435 int bam_write1(bamFile fp, const bam1_t *b);
438 @abstract Initiate a pointer to bam1_t struct
440 #define bam_init1() ((bam1_t*)calloc(1, sizeof(bam1_t)))
443 @abstract Free the memory allocated for an alignment.
444 @param b pointer to an alignment
446 #define bam_destroy1(b) do { \
447 if (b) { free((b)->data); free(b); } \
451 @abstract Format a BAM record in the SAM format
452 @param header pointer to the header structure
453 @param b alignment to print
454 @return a pointer to the SAM string
456 char *bam_format1(const bam_header_t *header, const bam1_t *b);
458 char *bam_format1_core(const bam_header_t *header, const bam1_t *b, int of);
461 @abstract Check whether a BAM record is plausibly valid
462 @param header associated header structure, or NULL if unavailable
463 @param b alignment to validate
464 @return 0 if the alignment is invalid; non-zero otherwise
466 @discussion Simple consistency check of some of the fields of the
467 alignment record. If the header is provided, several additional checks
468 are made. Not all fields are checked, so a non-zero result is not a
469 guarantee that the record is valid. However it is usually good enough
470 to detect when bam_seek() has been called with a virtual file offset
471 that is not the offset of an alignment record.
473 int bam_validate1(const bam_header_t *header, const bam1_t *b);
475 const char *bam_get_library(bam_header_t *header, const bam1_t *b);
483 @abstract Structure for one alignment covering the pileup position.
484 @field b pointer to the alignment
485 @field qpos position of the read base at the pileup site, 0-based
486 @field indel indel length; 0 for no indel, positive for ins and negative for del
487 @field is_del 1 iff the base on the padded read is a deletion
488 @field level the level of the read in the "viewer" mode
490 @discussion See also bam_plbuf_push() and bam_lplbuf_push(). The
491 difference between the two functions is that the former does not
492 set bam_pileup1_t::level, while the later does. Level helps the
493 implementation of alignment viewers, but calculating this has some
500 uint32_t is_del:1, is_head:1, is_tail:1, is_refskip:1, aux:28;
503 typedef int (*bam_plp_auto_f)(void *data, bam1_t *b);
506 typedef struct __bam_plp_t *bam_plp_t;
508 bam_plp_t bam_plp_init(bam_plp_auto_f func, void *data);
509 int bam_plp_push(bam_plp_t iter, const bam1_t *b);
510 const bam_pileup1_t *bam_plp_next(bam_plp_t iter, int *_tid, int *_pos, int *_n_plp);
511 const bam_pileup1_t *bam_plp_auto(bam_plp_t iter, int *_tid, int *_pos, int *_n_plp);
512 void bam_plp_set_mask(bam_plp_t iter, int mask);
513 void bam_plp_set_maxcnt(bam_plp_t iter, int maxcnt);
514 void bam_plp_reset(bam_plp_t iter);
515 void bam_plp_destroy(bam_plp_t iter);
518 typedef struct __bam_mplp_t *bam_mplp_t;
520 bam_mplp_t bam_mplp_init(int n, bam_plp_auto_f func, void **data);
521 void bam_mplp_destroy(bam_mplp_t iter);
522 void bam_mplp_set_maxcnt(bam_mplp_t iter, int maxcnt);
523 int bam_mplp_auto(bam_mplp_t iter, int *_tid, int *_pos, int *n_plp, const bam_pileup1_t **plp);
526 @abstract Type of function to be called by bam_plbuf_push().
527 @param tid chromosome ID as is defined in the header
528 @param pos start coordinate of the alignment, 0-based
529 @param n number of elements in pl array
530 @param pl array of alignments
531 @param data user provided data
532 @discussion See also bam_plbuf_push(), bam_plbuf_init() and bam_pileup1_t.
534 typedef int (*bam_pileup_f)(uint32_t tid, uint32_t pos, int n, const bam_pileup1_t *pl, void *data);
542 void bam_plbuf_set_mask(bam_plbuf_t *buf, int mask);
543 void bam_plbuf_reset(bam_plbuf_t *buf);
544 bam_plbuf_t *bam_plbuf_init(bam_pileup_f func, void *data);
545 void bam_plbuf_destroy(bam_plbuf_t *buf);
546 int bam_plbuf_push(const bam1_t *b, bam_plbuf_t *buf);
548 int bam_pileup_file(bamFile fp, int mask, bam_pileup_f func, void *func_data);
550 struct __bam_lplbuf_t;
551 typedef struct __bam_lplbuf_t bam_lplbuf_t;
553 void bam_lplbuf_reset(bam_lplbuf_t *buf);
555 /*! @abstract bam_plbuf_init() equivalent with level calculated. */
556 bam_lplbuf_t *bam_lplbuf_init(bam_pileup_f func, void *data);
558 /*! @abstract bam_plbuf_destroy() equivalent with level calculated. */
559 void bam_lplbuf_destroy(bam_lplbuf_t *tv);
561 /*! @abstract bam_plbuf_push() equivalent with level calculated. */
562 int bam_lplbuf_push(const bam1_t *b, bam_lplbuf_t *buf);
565 /*********************
566 * BAM indexing APIs *
567 *********************/
569 struct __bam_index_t;
570 typedef struct __bam_index_t bam_index_t;
573 @abstract Build index for a BAM file.
574 @discussion Index file "fn.bai" will be created.
575 @param fn name of the BAM file
576 @return always 0 currently
578 int bam_index_build(const char *fn);
581 @abstract Load index from file "fn.bai".
582 @param fn name of the BAM file (NOT the index file)
583 @return pointer to the index structure
585 bam_index_t *bam_index_load(const char *fn);
588 @abstract Destroy an index structure.
589 @param idx pointer to the index structure
591 void bam_index_destroy(bam_index_t *idx);
594 @abstract Type of function to be called by bam_fetch().
595 @param b the alignment
596 @param data user provided data
598 typedef int (*bam_fetch_f)(const bam1_t *b, void *data);
601 @abstract Retrieve the alignments that are overlapped with the
604 @discussion A user defined function will be called for each
605 retrieved alignment ordered by its start position.
607 @param fp BAM file handler
608 @param idx pointer to the alignment index
609 @param tid chromosome ID as is defined in the header
610 @param beg start coordinate, 0-based
611 @param end end coordinate, 0-based
612 @param data user provided data (will be transferred to func)
613 @param func user defined function
615 int bam_fetch(bamFile fp, const bam_index_t *idx, int tid, int beg, int end, void *data, bam_fetch_f func);
617 bam_iter_t bam_iter_query(const bam_index_t *idx, int tid, int beg, int end);
618 int bam_iter_read(bamFile fp, bam_iter_t iter, bam1_t *b);
619 void bam_iter_destroy(bam_iter_t iter);
622 @abstract Parse a region in the format: "chr2:100,000-200,000".
623 @discussion bam_header_t::hash will be initialized if empty.
624 @param header pointer to the header structure
625 @param str string to be parsed
626 @param ref_id the returned chromosome ID
627 @param begin the returned start coordinate
628 @param end the returned end coordinate
629 @return 0 on success; -1 on failure
631 int bam_parse_region(bam_header_t *header, const char *str, int *ref_id, int *begin, int *end);
634 /**************************
635 * APIs for optional tags *
636 **************************/
639 @abstract Retrieve data of a tag
640 @param b pointer to an alignment struct
641 @param tag two-character tag to be retrieved
643 @return pointer to the type and data. The first character is the
644 type that can be 'iIsScCdfAZH'.
646 @discussion Use bam_aux2?() series to convert the returned data to
647 the corresponding type.
649 uint8_t *bam_aux_get(const bam1_t *b, const char tag[2]);
651 int32_t bam_aux2i(const uint8_t *s);
652 float bam_aux2f(const uint8_t *s);
653 double bam_aux2d(const uint8_t *s);
654 char bam_aux2A(const uint8_t *s);
655 char *bam_aux2Z(const uint8_t *s);
657 int bam_aux_del(bam1_t *b, uint8_t *s);
658 void bam_aux_append(bam1_t *b, const char tag[2], char type, int len, uint8_t *data);
659 uint8_t *bam_aux_get_core(bam1_t *b, const char tag[2]); // an alias of bam_aux_get()
667 @abstract Calculate the rightmost coordinate of an alignment on the
670 @param c pointer to the bam1_core_t structure
671 @param cigar the corresponding CIGAR array (from bam1_t::cigar)
672 @return the rightmost coordinate, 0-based
674 uint32_t bam_calend(const bam1_core_t *c, const uint32_t *cigar);
677 @abstract Calculate the length of the query sequence from CIGAR.
678 @param c pointer to the bam1_core_t structure
679 @param cigar the corresponding CIGAR array (from bam1_t::cigar)
680 @return length of the query sequence
682 int32_t bam_cigar2qlen(const bam1_core_t *c, const uint32_t *cigar);
689 @abstract Calculate the minimum bin that contains a region [beg,end).
690 @param beg start of the region, 0-based
691 @param end end of the region, 0-based
694 static inline int bam_reg2bin(uint32_t beg, uint32_t end)
697 if (beg>>14 == end>>14) return 4681 + (beg>>14);
698 if (beg>>17 == end>>17) return 585 + (beg>>17);
699 if (beg>>20 == end>>20) return 73 + (beg>>20);
700 if (beg>>23 == end>>23) return 9 + (beg>>23);
701 if (beg>>26 == end>>26) return 1 + (beg>>26);
706 @abstract Copy an alignment
707 @param bdst destination alignment struct
708 @param bsrc source alignment struct
709 @return pointer to the destination alignment struct
711 static inline bam1_t *bam_copy1(bam1_t *bdst, const bam1_t *bsrc)
713 uint8_t *data = bdst->data;
714 int m_data = bdst->m_data; // backup data and m_data
715 if (m_data < bsrc->data_len) { // double the capacity
716 m_data = bsrc->data_len; kroundup32(m_data);
717 data = (uint8_t*)realloc(data, m_data);
719 memcpy(data, bsrc->data, bsrc->data_len); // copy var-len data
720 *bdst = *bsrc; // copy the rest
721 // restore the backup
722 bdst->m_data = m_data;
728 @abstract Duplicate an alignment
729 @param src source alignment struct
730 @return pointer to the destination alignment struct
732 static inline bam1_t *bam_dup1(const bam1_t *src)
737 b->m_data = b->data_len;
738 b->data = (uint8_t*)calloc(b->data_len, 1);
739 memcpy(b->data, src->data, b->data_len);