3 Copyright (c) 2008 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.
51 #if _IOLIB == 1 && !defined(_NO_RAZF)
52 #define BAM_TRUE_OFFSET
54 /*! @abstract BAM file handler */
55 typedef RAZF *bamFile;
56 #define bam_open(fn, mode) razf_open(fn, mode)
57 #define bam_dopen(fd, mode) razf_dopen(fd, mode)
58 #define bam_close(fp) razf_close(fp)
59 #define bam_read(fp, buf, size) razf_read(fp, buf, size)
60 #define bam_write(fp, buf, size) razf_write(fp, buf, size)
61 #define bam_tell(fp) razf_tell(fp)
62 #define bam_seek(fp, pos, dir) razf_seek(fp, pos, dir)
64 #define BAM_VIRTUAL_OFFSET16
66 /*! @abstract BAM file handler */
67 typedef BGZF *bamFile;
68 #define bam_open(fn, mode) bgzf_open(fn, mode)
69 #define bam_dopen(fd, mode) bgzf_fdopen(fd, mode)
70 #define bam_close(fp) bgzf_close(fp)
71 #define bam_read(fp, buf, size) bgzf_read(fp, buf, size)
72 #define bam_write(fp, buf, size) bgzf_write(fp, buf, size)
73 #define bam_tell(fp) bgzf_tell(fp)
74 #define bam_seek(fp, pos, dir) bgzf_seek(fp, pos, dir)
76 #define BAM_VIRTUAL_OFFSET16
78 /*! @abstract BAM file handler */
79 typedef RAZF *bamFile;
80 #define bam_open(fn, mode) razf_open2(fn, mode)
81 #define bam_dopen(fd, mode) razf_dopen2(fd, mode)
82 #define bam_close(fp) razf_close(fp)
83 #define bam_read(fp, buf, size) razf_read(fp, buf, size)
84 #define bam_write(fp, buf, size) razf_write(fp, buf, size)
85 #define bam_tell(fp) razf_tell2(fp)
86 #define bam_seek(fp, pos, dir) razf_seek2(fp, pos, dir)
90 @abstract Structure for the alignment header.
91 @field n_targets number of reference sequences
92 @field target_name names of the reference sequences
93 @field target_len lengths of the referene sequences
94 @field hash hash table for fast name lookup
95 @field rg2lib hash table for @RG-ID -> LB lookup
96 @field l_text length of the plain text in the header
97 @field text plain text
99 @discussion Field hash points to null by default. It is a private
105 uint32_t *target_len;
111 /*! @abstract the read is paired in sequencing, no matter whether it is mapped in a pair */
112 #define BAM_FPAIRED 1
113 /*! @abstract the read is mapped in a proper pair */
114 #define BAM_FPROPER_PAIR 2
115 /*! @abstract the read itself is unmapped; conflictive with BAM_FPROPER_PAIR */
117 /*! @abstract the mate is unmapped */
118 #define BAM_FMUNMAP 8
119 /*! @abstract the read is mapped to the reverse strand */
120 #define BAM_FREVERSE 16
121 /*! @abstract the mate is mapped to the reverse strand */
122 #define BAM_FMREVERSE 32
123 /*! @abstract this is read1 */
124 #define BAM_FREAD1 64
125 /*! @abstract this is read2 */
126 #define BAM_FREAD2 128
127 /*! @abstract not primary alignment */
128 #define BAM_FSECONDARY 256
129 /*! @abstract QC failure */
130 #define BAM_FQCFAIL 512
131 /*! @abstract optical or PCR duplicate */
132 #define BAM_FDUP 1024
134 /*! @abstract defautl mask for pileup */
135 #define BAM_DEF_MASK (BAM_FUNMAP | BAM_FSECONDARY | BAM_FQCFAIL | BAM_FDUP)
137 #define BAM_CORE_SIZE sizeof(bam1_core_t)
140 * Describing how CIGAR operation/length is packed in a 32-bit integer.
142 #define BAM_CIGAR_SHIFT 4
143 #define BAM_CIGAR_MASK ((1 << BAM_CIGAR_SHIFT) - 1)
148 /*! @abstract CIGAR: match */
150 /*! @abstract CIGAR: insertion to the reference */
152 /*! @abstract CIGAR: deletion from the reference */
154 /*! @abstract CIGAR: skip on the reference (e.g. spliced alignment) */
155 #define BAM_CREF_SKIP 3
156 /*! @abstract CIGAR: clip on the read with clipped sequence present in qseq */
157 #define BAM_CSOFT_CLIP 4
158 /*! @abstract CIGAR: clip on the read with clipped sequence trimmed off */
159 #define BAM_CHARD_CLIP 5
160 /*! @abstract CIGAR: padding */
164 @abstract Structure for core alignment information.
165 @field tid chromosome ID, defined by bam_header_t
166 @field pos 0-based leftmost coordinate
167 @field strand strand; 0 for forward and 1 otherwise
168 @field bin bin calculated by bam_reg2bin()
169 @field qual mapping quality
170 @field l_qname length of the query name
171 @field flag bitwise flag
172 @field n_cigar number of CIGAR operations
173 @field l_qseq length of the query sequence (read)
178 uint32_t bin:16, qual:8, l_qname:8;
179 uint32_t flag:16, n_cigar:16;
187 @abstract Structure for one alignment.
188 @field core core information about the alignment
189 @field l_aux length of auxiliary data
190 @field data_len current length of bam1_t::data
191 @field m_data maximum length of bam1_t::data
192 @field data all variable-length data, concatenated; structure: cigar-qname-seq-qual-aux
196 1. qname is zero tailing and core.l_qname includes the tailing '\0'.
197 2. l_qseq is calculated from the total length of an alignment block
198 on reading or from CIGAR.
202 int l_aux, data_len, m_data;
206 #define bam1_strand(b) (((b)->core.flag&BAM_FREVERSE) != 0)
207 #define bam1_mstrand(b) (((b)->core.flag&BAM_FMREVERSE) != 0)
210 @abstract Get the CIGAR array
211 @param b pointer to an alignment
212 @return pointer to the CIGAR array
214 @discussion In the CIGAR array, each element is a 32-bit integer. The
215 lower 4 bits gives a CIGAR operation and the higher 28 bits keep the
218 #define bam1_cigar(b) ((uint32_t*)((b)->data + (b)->core.l_qname))
221 @abstract Get the name of the query
222 @param b pointer to an alignment
223 @return pointer to the name string, null terminated
225 #define bam1_qname(b) ((char*)((b)->data))
228 @abstract Get query sequence
229 @param b pointer to an alignment
230 @return pointer to sequence
232 @discussion Each base is encoded in 4 bits: 1 for A, 2 for C, 4 for G,
233 8 for T and 15 for N. Two bases are packed in one byte with the base
234 at the higher 4 bits having smaller coordinate on the read. It is
235 recommended to use bam1_seqi() macro to get the base.
237 #define bam1_seq(b) ((b)->data + (b)->core.n_cigar*4 + (b)->core.l_qname)
240 @abstract Get query quality
241 @param b pointer to an alignment
242 @return pointer to quality string
244 #define bam1_qual(b) ((b)->data + (b)->core.n_cigar*4 + (b)->core.l_qname + ((b)->core.l_qseq + 1)/2)
247 @abstract Get a base on read
248 @param s Query sequence returned by bam1_seq()
249 @param i The i-th position, 0-based
250 @return 4-bit integer representing the base.
252 #define bam1_seqi(s, i) ((s)[(i)/2] >> 4*(1-(i)%2) & 0xf)
255 @abstract Get query sequence and quality
256 @param b pointer to an alignment
257 @return pointer to the concatenated auxiliary data
259 #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)
263 @abstract Round an integer to the next closest power-2 integer.
264 @param x integer to be rounded (in place)
265 @discussion x will be modified.
267 #define kroundup32(x) (--(x), (x)|=(x)>>1, (x)|=(x)>>2, (x)|=(x)>>4, (x)|=(x)>>8, (x)|=(x)>>16, ++(x))
271 @abstract Whether the machine is big-endian; modified only in
274 extern int bam_is_be;
276 /*! @abstract Table for converting a nucleotide character to the 4-bit encoding. */
277 extern unsigned char bam_nt16_table[256];
279 /*! @abstract Table for converting a 4-bit encoded nucleotide to a letter. */
280 extern char *bam_nt16_rev_table;
282 extern char bam_nt16_nt4_table[];
288 /*! @abstract TAM file handler */
289 typedef struct __tamFile_t *tamFile;
292 @abstract Open a SAM file for reading, either uncompressed or compressed by gzip/zlib.
293 @param fn SAM file name
294 @return SAM file handler
296 tamFile sam_open(const char *fn);
299 @abstract Close a SAM file handler
300 @param fp SAM file handler
302 void sam_close(tamFile fp);
305 @abstract Read one alignment from a SAM file handler
306 @param fp SAM file handler
307 @param header header information (ordered names of chromosomes)
308 @param b read alignment; all members in b will be updated
309 @return 0 if successful; otherwise negative
311 int sam_read1(tamFile fp, bam_header_t *header, bam1_t *b);
314 @abstract Read header information from a TAB-delimited list file.
315 @param fn_list file name for the list
316 @return a pointer to the header structure
318 @discussion Each line in this file consists of chromosome name and
319 the length of chromosome.
321 bam_header_t *sam_header_read2(const char *fn_list);
324 @abstract Read header from a SAM file (if present)
325 @param fp SAM file handler
326 @return pointer to header struct; 0 if no @SQ lines available
328 bam_header_t *sam_header_read(tamFile fp);
331 @abstract Parse @SQ lines a update a header struct
332 @param h pointer to the header struct to be updated
333 @return number of target sequences
335 @discussion bam_header_t::{n_targets,target_len,target_name} will
336 be destroyed in the first place.
338 int sam_header_parse(bam_header_t *h);
341 @abstract Parse @RG lines a update a header struct
342 @param h pointer to the header struct to be updated
343 @return number of @RG lines
345 @discussion bam_header_t::rg2lib will be destroyed in the first
348 int sam_header_parse_rg(bam_header_t *h);
350 #define sam_write1(header, b) bam_view1(header, b)
352 int bam_strmap_put(void *strmap, const char *rg, const char *lib);
353 const char *bam_strmap_get(const void *strmap, const char *rg);
354 void *bam_strmap_dup(const void*);
355 void *bam_strmap_init();
356 void bam_strmap_destroy(void *strmap);
359 @abstract Initialize a header structure.
360 @return the pointer to the header structure
362 @discussion This function also modifies the global variable
365 bam_header_t *bam_header_init();
368 @abstract Destroy a header structure.
369 @param header pointer to the header
371 void bam_header_destroy(bam_header_t *header);
374 @abstract Read a header structure from BAM.
375 @param fp BAM file handler, opened by bam_open()
376 @return pointer to the header structure
378 @discussion The file position indicator must be placed at the
379 beginning of the file. Upon success, the position indicator will
380 be set at the start of the first alignment.
382 bam_header_t *bam_header_read(bamFile fp);
385 @abstract Write a header structure to BAM.
386 @param fp BAM file handler
387 @param header pointer to the header structure
388 @return always 0 currently
390 int bam_header_write(bamFile fp, const bam_header_t *header);
393 @abstract Read an alignment from BAM.
394 @param fp BAM file handler
395 @param b read alignment; all members are updated.
396 @return number of bytes read from the file
398 @discussion The file position indicator must be
399 placed right before an alignment. Upon success, this function
400 will set the position indicator to the start of the next
401 alignment. This function is not affected by the machine
404 int bam_read1(bamFile fp, bam1_t *b);
407 @abstract Write an alignment to BAM.
408 @param fp BAM file handler
409 @param c pointer to the bam1_core_t structure
410 @param data_len total length of variable size data related to
412 @param data pointer to the concatenated data
413 @return number of bytes written to the file
415 @discussion This function is not affected by the machine
418 int bam_write1_core(bamFile fp, const bam1_core_t *c, int data_len, uint8_t *data);
421 @abstract Write an alignment to BAM.
422 @param fp BAM file handler
423 @param b alignment to write
424 @return number of bytes written to the file
426 @abstract It is equivalent to:
427 bam_write1_core(fp, &b->core, b->data_len, b->data)
429 int bam_write1(bamFile fp, const bam1_t *b);
432 @abstract Initiate a pointer to bam1_t struct
434 #define bam_init1() ((bam1_t*)calloc(1, sizeof(bam1_t)))
437 @abstract Free the memory allocated for an alignment.
438 @param b pointer to an alignment
440 #define bam_destroy1(b) do { \
441 free((b)->data); free(b); \
445 @abstract Format a BAM record in the SAM format
446 @param header pointer to the header structure
447 @param b alignment to print
448 @return a pointer to the SAM string
450 char *bam_format1(const bam_header_t *header, const bam1_t *b);
453 @abstract Structure for one alignment covering the pileup position.
454 @field b pointer to the alignment
455 @field qpos position of the read base at the pileup site, 0-based
456 @field indel indel length; 0 for no indel, positive for ins and negative for del
457 @field is_del 1 iff the base on the padded read is a deletion
458 @field level the level of the read in the "viewer" mode
460 @discussion See also bam_plbuf_push() and bam_lplbuf_push(). The
461 difference between the two functions is that the former does not
462 set bam_pileup1_t::level, while the later does. Level helps the
463 implementation of alignment viewers, but calculating this has some
470 uint32_t is_del:1, is_head:1, is_tail:1;
473 struct __bam_plbuf_t;
474 /*! @abstract pileup buffer */
475 typedef struct __bam_plbuf_t bam_plbuf_t;
477 void bam_plbuf_set_mask(bam_plbuf_t *buf, int mask);
480 @abstract Type of function to be called by bam_plbuf_push().
481 @param tid chromosome ID as is defined in the header
482 @param pos start coordinate of the alignment, 0-based
483 @param n number of elements in pl array
484 @param pl array of alignments
485 @param data user provided data
486 @discussion See also bam_plbuf_push(), bam_plbuf_init() and bam_pileup1_t.
488 typedef int (*bam_pileup_f)(uint32_t tid, uint32_t pos, int n, const bam_pileup1_t *pl, void *data);
491 @abstract Reset a pileup buffer for another pileup process
492 @param buf the pileup buffer to be reset
494 void bam_plbuf_reset(bam_plbuf_t *buf);
497 @abstract Initialize a buffer for pileup.
498 @param func fucntion to be called by bam_pileup_core()
499 @param data user provided data
500 @return pointer to the pileup buffer
502 bam_plbuf_t *bam_plbuf_init(bam_pileup_f func, void *data);
505 @abstract Destroy a pileup buffer.
506 @param buf pointer to the pileup buffer
508 void bam_plbuf_destroy(bam_plbuf_t *buf);
511 @abstract Push an alignment to the pileup buffer.
512 @param b alignment to be pushed
513 @param buf pileup buffer
514 @see bam_plbuf_init()
515 @return always 0 currently
517 @discussion If all the alignments covering a particular site have
518 been collected, this function will call the user defined function
519 as is provided to bam_plbuf_init(). The coordinate of the site and
520 all the alignments will be transferred to the user defined
521 function as function parameters.
523 When all the alignments are pushed to the buffer, this function
524 needs to be called with b equal to NULL. This will flush the
525 buffer. A pileup buffer can only be reused when bam_plbuf_reset()
528 int bam_plbuf_push(const bam1_t *b, bam_plbuf_t *buf);
530 struct __bam_lplbuf_t;
531 typedef struct __bam_lplbuf_t bam_lplbuf_t;
533 void bam_lplbuf_reset(bam_lplbuf_t *buf);
535 /*! @abstract bam_plbuf_init() equivalent with level calculated. */
536 bam_lplbuf_t *bam_lplbuf_init(bam_pileup_f func, void *data);
538 /*! @abstract bam_plbuf_destroy() equivalent with level calculated. */
539 void bam_lplbuf_destroy(bam_lplbuf_t *tv);
541 /*! @abstract bam_plbuf_push() equivalent with level calculated. */
542 int bam_lplbuf_push(const bam1_t *b, bam_lplbuf_t *buf);
544 /*! @abstract bam_plbuf_file() equivalent with level calculated. */
545 int bam_lpileup_file(bamFile fp, int mask, bam_pileup_f func, void *func_data);
547 struct __bam_index_t;
548 typedef struct __bam_index_t bam_index_t;
551 @abstract Build index for a BAM file.
552 @discussion Index file "fn.bai" will be created.
553 @param fn name of the BAM file
554 @return always 0 currently
556 int bam_index_build(const char *fn);
559 @abstract Load index from file "fn.bai".
560 @param fn name of the BAM file (NOT the index file)
561 @return pointer to the index structure
563 bam_index_t *bam_index_load(const char *fn);
566 @abstract Destroy an index structure.
567 @param idx pointer to the index structure
569 void bam_index_destroy(bam_index_t *idx);
572 @abstract Type of function to be called by bam_fetch().
573 @param b the alignment
574 @param data user provided data
576 typedef int (*bam_fetch_f)(const bam1_t *b, void *data);
579 @abstract Retrieve the alignments that are overlapped with the
582 @discussion A user defined function will be called for each
583 retrieved alignment ordered by its start position.
585 @param fp BAM file handler
586 @param idx pointer to the alignment index
587 @param tid chromosome ID as is defined in the header
588 @param beg start coordinate, 0-based
589 @param end end coordinate, 0-based
590 @param data user provided data (will be transferred to func)
591 @param func user defined function
593 int bam_fetch(bamFile fp, const bam_index_t *idx, int tid, int beg, int end, void *data, bam_fetch_f func);
596 @abstract Parse a region in the format: "chr2:100,000-200,000".
597 @discussion bam_header_t::hash will be initialized if empty.
598 @param header pointer to the header structure
599 @param str string to be parsed
600 @param ref_id the returned chromosome ID
601 @param begin the returned start coordinate
602 @param end the returned end coordinate
604 void bam_parse_region(bam_header_t *header, const char *str, int *ref_id, int *begin, int *end);
607 @abstract Retrieve data of a tag
608 @param b pointer to an alignment struct
609 @param tag two-character tag to be retrieved
611 @return pointer to the type and data. The first character is the
612 type that can be 'iIsScCdfAZH'.
614 @discussion Use bam_aux2?() series to convert the returned data to
615 the corresponding type.
617 uint8_t *bam_aux_get(const bam1_t *b, const char tag[2]);
619 int32_t bam_aux2i(const uint8_t *s);
620 float bam_aux2f(const uint8_t *s);
621 double bam_aux2d(const uint8_t *s);
622 char bam_aux2A(const uint8_t *s);
623 char *bam_aux2Z(const uint8_t *s);
625 void bam_aux_append(bam1_t *b, const char tag[2], char type, int len, uint8_t *data);
627 // uint8_t *bam_aux_get_core(bam1_t *b, const char tag[2]); // an alias of bam_aux_get()
630 @abstract Calculate the rightmost coordinate of an alignment on the
633 @param c pointer to the bam1_core_t structure
634 @param cigar the corresponding CIGAR array (from bam1_t::cigar)
635 @return the rightmost coordinate, 0-based
637 uint32_t bam_calend(const bam1_core_t *c, const uint32_t *cigar);
640 @abstract Calculate the length of the query sequence from CIGAR.
641 @param c pointer to the bam1_core_t structure
642 @param cigar the corresponding CIGAR array (from bam1_t::cigar)
643 @return length of the query sequence
645 int32_t bam_cigar2qlen(const bam1_core_t *c, const uint32_t *cigar);
653 int bam_segreg(int32_t pos, const bam1_core_t *c, const uint32_t *cigar, bam_segreg_t *reg);
660 @abstract Calculate the minimum bin that contains a region [beg,end).
661 @param beg start of the region, 0-based
662 @param end end of the region, 0-based
665 static inline int bam_reg2bin(uint32_t beg, uint32_t end)
668 if (beg>>14 == end>>14) return 4681 + (beg>>14);
669 if (beg>>17 == end>>17) return 585 + (beg>>17);
670 if (beg>>20 == end>>20) return 73 + (beg>>20);
671 if (beg>>23 == end>>23) return 9 + (beg>>23);
672 if (beg>>26 == end>>26) return 1 + (beg>>26);
677 @abstract Copy an alignment
678 @param bdst destination alignment struct
679 @param bsrc source alignment struct
680 @return pointer to the destination alignment struct
682 static inline bam1_t *bam_copy1(bam1_t *bdst, const bam1_t *bsrc)
684 uint8_t *data = bdst->data;
685 int m_data = bdst->m_data; // backup data and m_data
686 if (m_data < bsrc->m_data) { // double the capacity
687 m_data = bsrc->m_data; kroundup32(m_data);
688 data = (uint8_t*)realloc(data, m_data);
690 memcpy(data, bsrc->data, bsrc->data_len); // copy var-len data
691 *bdst = *bsrc; // copy the rest
692 // restore the backup
693 bdst->m_data = m_data;
699 @abstract Duplicate an alignment
700 @param src source alignment struct
701 @return pointer to the destination alignment struct
703 static inline bam1_t *bam_dup1(const bam1_t *src)
708 b->m_data = b->data_len;
709 b->data = (uint8_t*)calloc(b->data_len, 1);
710 memcpy(b->data, src->data, b->data_len);