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.
50 #if _IOLIB == 1 && !defined(_NO_RAZF)
51 #define BAM_TRUE_OFFSET
53 /*! @abstract BAM file handler */
54 typedef RAZF *bamFile;
55 #define bam_open(fn, mode) razf_open(fn, mode)
56 #define bam_dopen(fd, mode) razf_dopen(fd, mode)
57 #define bam_close(fp) razf_close(fp)
58 #define bam_read(fp, buf, size) razf_read(fp, buf, size)
59 #define bam_write(fp, buf, size) razf_write(fp, buf, size)
60 #define bam_tell(fp) razf_tell(fp)
61 #define bam_seek(fp, pos, dir) razf_seek(fp, pos, dir)
63 #define BAM_VIRTUAL_OFFSET16
65 /*! @abstract BAM file handler */
66 typedef BGZF *bamFile;
67 #define bam_open(fn, mode) bgzf_open(fn, mode)
68 #define bam_dopen(fd, mode) bgzf_fdopen(fd, mode)
69 #define bam_close(fp) bgzf_close(fp)
70 #define bam_read(fp, buf, size) bgzf_read(fp, buf, size)
71 #define bam_write(fp, buf, size) bgzf_write(fp, buf, size)
72 #define bam_tell(fp) bgzf_tell(fp)
73 #define bam_seek(fp, pos, dir) bgzf_seek(fp, pos, dir)
75 #define BAM_VIRTUAL_OFFSET16
77 /*! @abstract BAM file handler */
78 typedef RAZF *bamFile;
79 #define bam_open(fn, mode) razf_open2(fn, mode)
80 #define bam_dopen(fd, mode) razf_dopen2(fd, mode)
81 #define bam_close(fp) razf_close(fp)
82 #define bam_read(fp, buf, size) razf_read(fp, buf, size)
83 #define bam_write(fp, buf, size) razf_write(fp, buf, size)
84 #define bam_tell(fp) razf_tell2(fp)
85 #define bam_seek(fp, pos, dir) razf_seek2(fp, pos, dir)
89 @abstract Structure for the alignment header.
90 @field n_targets number of reference sequences
91 @field target_name names of the reference sequences
92 @field target_len lengths of the referene sequences
93 @field hash hash table for fast name lookup
94 @field rg2lib hash table for @RG-ID -> LB lookup
95 @field l_text length of the plain text in the header
96 @field text plain text
98 @discussion Field hash points to null by default. It is a private
104 uint32_t *target_len;
110 /*! @abstract the read is paired in sequencing, no matter whether it is mapped in a pair */
111 #define BAM_FPAIRED 1
112 /*! @abstract the read is mapped in a proper pair */
113 #define BAM_FPROPER_PAIR 2
114 /*! @abstract the read itself is unmapped; conflictive with BAM_FPROPER_PAIR */
116 /*! @abstract the mate is unmapped */
117 #define BAM_FMUNMAP 8
118 /*! @abstract the read is mapped to the reverse strand */
119 #define BAM_FREVERSE 16
120 /*! @abstract the mate is mapped to the reverse strand */
121 #define BAM_FMREVERSE 32
122 /*! @abstract this is read1 */
123 #define BAM_FREAD1 64
124 /*! @abstract this is read2 */
125 #define BAM_FREAD2 128
126 /*! @abstract not primary alignment */
127 #define BAM_FSECONDARY 256
128 /*! @abstract QC failure */
129 #define BAM_FQCFAIL 512
130 /*! @abstract optical or PCR duplicate */
131 #define BAM_FDUP 1024
133 /*! @abstract defautl mask for pileup */
134 #define BAM_DEF_MASK (BAM_FUNMAP | BAM_FSECONDARY | BAM_FQCFAIL | BAM_FDUP)
136 #define BAM_CORE_SIZE sizeof(bam1_core_t)
139 * Describing how CIGAR operation/length is packed in a 32-bit integer.
141 #define BAM_CIGAR_SHIFT 4
142 #define BAM_CIGAR_MASK ((1 << BAM_CIGAR_SHIFT) - 1)
147 /*! @abstract CIGAR: match */
149 /*! @abstract CIGAR: insertion to the reference */
151 /*! @abstract CIGAR: deletion from the reference */
153 /*! @abstract CIGAR: skip on the reference (e.g. spliced alignment) */
154 #define BAM_CREF_SKIP 3
155 /*! @abstract CIGAR: clip on the read with clipped sequence present in qseq */
156 #define BAM_CSOFT_CLIP 4
157 /*! @abstract CIGAR: clip on the read with clipped sequence trimmed off */
158 #define BAM_CHARD_CLIP 5
159 /*! @abstract CIGAR: padding */
163 @abstract Structure for core alignment information.
164 @field tid chromosome ID, defined by bam_header_t
165 @field pos 0-based leftmost coordinate
166 @field strand strand; 0 for forward and 1 otherwise
167 @field bin bin calculated by bam_reg2bin()
168 @field qual mapping quality
169 @field l_qname length of the query name
170 @field flag bitwise flag
171 @field n_cigar number of CIGAR operations
172 @field l_qseq length of the query sequence (read)
177 uint32_t bin:16, qual:8, l_qname:8;
178 uint32_t flag:16, n_cigar:16;
186 @abstract Structure for one alignment.
187 @field core core information about the alignment
188 @field l_aux length of auxiliary data
189 @field data_len current length of bam1_t::data
190 @field m_data maximum length of bam1_t::data
191 @field data all variable-length data, concatenated; structure: cigar-qname-seq-qual-aux
195 1. qname is zero tailing and core.l_qname includes the tailing '\0'.
196 2. l_qseq is calculated from the total length of an alignment block
197 on reading or from CIGAR.
201 int l_aux, data_len, m_data;
205 #define bam1_strand(b) (((b)->core.flag&BAM_FREVERSE) != 0)
206 #define bam1_mstrand(b) (((b)->core.flag&BAM_FMREVERSE) != 0)
209 @abstract Get the CIGAR array
210 @param b pointer to an alignment
211 @return pointer to the CIGAR array
213 @discussion In the CIGAR array, each element is a 32-bit integer. The
214 lower 4 bits gives a CIGAR operation and the higher 28 bits keep the
217 #define bam1_cigar(b) ((uint32_t*)((b)->data + (b)->core.l_qname))
220 @abstract Get the name of the query
221 @param b pointer to an alignment
222 @return pointer to the name string, null terminated
224 #define bam1_qname(b) ((char*)((b)->data))
227 @abstract Get query sequence
228 @param b pointer to an alignment
229 @return pointer to sequence
231 @discussion Each base is encoded in 4 bits: 1 for A, 2 for C, 4 for G,
232 8 for T and 15 for N. Two bases are packed in one byte with the base
233 at the higher 4 bits having smaller coordinate on the read. It is
234 recommended to use bam1_seqi() macro to get the base.
236 #define bam1_seq(b) ((b)->data + (b)->core.n_cigar*4 + (b)->core.l_qname)
239 @abstract Get query quality
240 @param b pointer to an alignment
241 @return pointer to quality string
243 #define bam1_qual(b) ((b)->data + (b)->core.n_cigar*4 + (b)->core.l_qname + ((b)->core.l_qseq + 1)/2)
246 @abstract Get a base on read
247 @param s Query sequence returned by bam1_seq()
248 @param i The i-th position, 0-based
249 @return 4-bit integer representing the base.
251 #define bam1_seqi(s, i) ((s)[(i)/2] >> 4*(1-(i)%2) & 0xf)
254 @abstract Get query sequence and quality
255 @param b pointer to an alignment
256 @return pointer to the concatenated auxiliary data
258 #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)
262 @abstract Round an integer to the next closest power-2 integer.
263 @param x integer to be rounded (in place)
264 @discussion x will be modified.
266 #define kroundup32(x) (--(x), (x)|=(x)>>1, (x)|=(x)>>2, (x)|=(x)>>4, (x)|=(x)>>8, (x)|=(x)>>16, ++(x))
270 @abstract Whether the machine is big-endian; modified only in
273 extern int bam_is_be;
275 /*! @abstract Table for converting a nucleotide character to the 4-bit encoding. */
276 extern unsigned char bam_nt16_table[256];
278 /*! @abstract Table for converting a 4-bit encoded nucleotide to a letter. */
279 extern char *bam_nt16_rev_table;
281 extern char bam_nt16_nt4_table[];
287 /*! @abstract TAM file handler */
288 typedef struct __tamFile_t *tamFile;
291 @abstract Open a SAM file for reading, either uncompressed or compressed by gzip/zlib.
292 @param fn SAM file name
293 @return SAM file handler
295 tamFile sam_open(const char *fn);
298 @abstract Close a SAM file handler
299 @param fp SAM file handler
301 void sam_close(tamFile fp);
304 @abstract Read one alignment from a SAM file handler
305 @param fp SAM file handler
306 @param header header information (ordered names of chromosomes)
307 @param b read alignment; all members in b will be updated
308 @return 0 if successful; otherwise negative
310 int sam_read1(tamFile fp, bam_header_t *header, bam1_t *b);
313 @abstract Read header information from a TAB-delimited list file.
314 @param fn_list file name for the list
315 @return a pointer to the header structure
317 @discussion Each line in this file consists of chromosome name and
318 the length of chromosome.
320 bam_header_t *sam_header_read2(const char *fn_list);
323 @abstract Read header from a SAM file (if present)
324 @param fp SAM file handler
325 @return pointer to header struct; 0 if no @SQ lines available
327 bam_header_t *sam_header_read(tamFile fp);
330 @abstract Parse @SQ lines a update a header struct
331 @param h pointer to the header struct to be updated
332 @return number of target sequences
334 @discussion bam_header_t::{n_targets,target_len,target_name} will
335 be destroyed in the first place.
337 int sam_header_parse(bam_header_t *h);
340 @abstract Parse @RG lines a update a header struct
341 @param h pointer to the header struct to be updated
342 @return number of @RG lines
344 @discussion bam_header_t::rg2lib will be destroyed in the first
347 int sam_header_parse_rg(bam_header_t *h);
349 #define sam_write1(header, b) bam_view1(header, b)
351 int bam_strmap_put(void *strmap, const char *rg, const char *lib);
352 const char *bam_strmap_get(const void *strmap, const char *rg);
353 void *bam_strmap_dup(const void*);
354 void *bam_strmap_init();
355 void bam_strmap_destroy(void *strmap);
358 @abstract Initialize a header structure.
359 @return the pointer to the header structure
361 @discussion This function also modifies the global variable
364 bam_header_t *bam_header_init();
367 @abstract Destroy a header structure.
368 @param header pointer to the header
370 void bam_header_destroy(bam_header_t *header);
373 @abstract Read a header structure from BAM.
374 @param fp BAM file handler, opened by bam_open()
375 @return pointer to the header structure
377 @discussion The file position indicator must be placed at the
378 beginning of the file. Upon success, the position indicator will
379 be set at the start of the first alignment.
381 bam_header_t *bam_header_read(bamFile fp);
384 @abstract Write a header structure to BAM.
385 @param fp BAM file handler
386 @param header pointer to the header structure
387 @return always 0 currently
389 int bam_header_write(bamFile fp, const bam_header_t *header);
392 @abstract Read an alignment from BAM.
393 @param fp BAM file handler
394 @param b read alignment; all members are updated.
395 @return number of bytes read from the file
397 @discussion The file position indicator must be
398 placed right before an alignment. Upon success, this function
399 will set the position indicator to the start of the next
400 alignment. This function is not affected by the machine
403 int bam_read1(bamFile fp, bam1_t *b);
406 @abstract Write an alignment to BAM.
407 @param fp BAM file handler
408 @param c pointer to the bam1_core_t structure
409 @param data_len total length of variable size data related to
411 @param data pointer to the concatenated data
412 @return number of bytes written to the file
414 @discussion This function is not affected by the machine
417 int bam_write1_core(bamFile fp, const bam1_core_t *c, int data_len, uint8_t *data);
420 @abstract Write an alignment to BAM.
421 @param fp BAM file handler
422 @param b alignment to write
423 @return number of bytes written to the file
425 @abstract It is equivalent to:
426 bam_write1_core(fp, &b->core, b->data_len, b->data)
428 int bam_write1(bamFile fp, const bam1_t *b);
431 @abstract Initiate a pointer to bam1_t struct
433 #define bam_init1() ((bam1_t*)calloc(1, sizeof(bam1_t)))
436 @abstract Free the memory allocated for an alignment.
437 @param b pointer to an alignment
439 #define bam_destroy1(b) do { \
440 free((b)->data); free(b); \
444 @abstract Format a BAM record in the SAM format
445 @param header pointer to the header structure
446 @param b alignment to print
447 @return a pointer to the SAM string
449 char *bam_format1(const bam_header_t *header, const bam1_t *b);
452 @abstract Structure for one alignment covering the pileup position.
453 @field b pointer to the alignment
454 @field qpos position of the read base at the pileup site, 0-based
455 @field indel indel length; 0 for no indel, positive for ins and negative for del
456 @field is_del 1 iff the base on the padded read is a deletion
457 @field level the level of the read in the "viewer" mode
459 @discussion See also bam_plbuf_push() and bam_lplbuf_push(). The
460 difference between the two functions is that the former does not
461 set bam_pileup1_t::level, while the later does. Level helps the
462 implementation of alignment viewers, but calculating this has some
469 uint32_t is_del:1, is_head:1, is_tail:1;
472 struct __bam_plbuf_t;
473 /*! @abstract pileup buffer */
474 typedef struct __bam_plbuf_t bam_plbuf_t;
476 void bam_plbuf_set_mask(bam_plbuf_t *buf, int mask);
479 @abstract Type of function to be called by bam_plbuf_push().
480 @param tid chromosome ID as is defined in the header
481 @param pos start coordinate of the alignment, 0-based
482 @param n number of elements in pl array
483 @param pl array of alignments
484 @param data user provided data
485 @discussion See also bam_plbuf_push(), bam_plbuf_init() and bam_pileup1_t.
487 typedef int (*bam_pileup_f)(uint32_t tid, uint32_t pos, int n, const bam_pileup1_t *pl, void *data);
490 @abstract Reset a pileup buffer for another pileup process
491 @param buf the pileup buffer to be reset
493 void bam_plbuf_reset(bam_plbuf_t *buf);
496 @abstract Initialize a buffer for pileup.
497 @param func fucntion to be called by bam_pileup_core()
498 @param data user provided data
499 @return pointer to the pileup buffer
501 bam_plbuf_t *bam_plbuf_init(bam_pileup_f func, void *data);
504 @abstract Destroy a pileup buffer.
505 @param buf pointer to the pileup buffer
507 void bam_plbuf_destroy(bam_plbuf_t *buf);
510 @abstract Push an alignment to the pileup buffer.
511 @param b alignment to be pushed
512 @param buf pileup buffer
513 @see bam_plbuf_init()
514 @return always 0 currently
516 @discussion If all the alignments covering a particular site have
517 been collected, this function will call the user defined function
518 as is provided to bam_plbuf_init(). The coordinate of the site and
519 all the alignments will be transferred to the user defined
520 function as function parameters.
522 When all the alignments are pushed to the buffer, this function
523 needs to be called with b equal to NULL. This will flush the
524 buffer. A pileup buffer can only be reused when bam_plbuf_reset()
527 int bam_plbuf_push(const bam1_t *b, bam_plbuf_t *buf);
529 struct __bam_lplbuf_t;
530 typedef struct __bam_lplbuf_t bam_lplbuf_t;
532 void bam_lplbuf_reset(bam_lplbuf_t *buf);
534 /*! @abstract bam_plbuf_init() equivalent with level calculated. */
535 bam_lplbuf_t *bam_lplbuf_init(bam_pileup_f func, void *data);
537 /*! @abstract bam_plbuf_destroy() equivalent with level calculated. */
538 void bam_lplbuf_destroy(bam_lplbuf_t *tv);
540 /*! @abstract bam_plbuf_push() equivalent with level calculated. */
541 int bam_lplbuf_push(const bam1_t *b, bam_lplbuf_t *buf);
543 /*! @abstract bam_plbuf_file() equivalent with level calculated. */
544 int bam_lpileup_file(bamFile fp, int mask, bam_pileup_f func, void *func_data);
546 struct __bam_index_t;
547 typedef struct __bam_index_t bam_index_t;
550 @abstract Build index for a BAM file.
551 @discussion Index file "fn.bai" will be created.
552 @param fn name of the BAM file
553 @return always 0 currently
555 int bam_index_build(const char *fn);
558 @abstract Load index from file "fn.bai".
559 @param fn name of the BAM file (NOT the index file)
560 @return pointer to the index structure
562 bam_index_t *bam_index_load(const char *fn);
565 @abstract Destroy an index structure.
566 @param idx pointer to the index structure
568 void bam_index_destroy(bam_index_t *idx);
571 @abstract Type of function to be called by bam_fetch().
572 @param b the alignment
573 @param data user provided data
575 typedef int (*bam_fetch_f)(const bam1_t *b, void *data);
578 @abstract Retrieve the alignments that are overlapped with the
581 @discussion A user defined function will be called for each
582 retrieved alignment ordered by its start position.
584 @param fp BAM file handler
585 @param idx pointer to the alignment index
586 @param tid chromosome ID as is defined in the header
587 @param beg start coordinate, 0-based
588 @param end end coordinate, 0-based
589 @param data user provided data (will be transferred to func)
590 @param func user defined function
592 int bam_fetch(bamFile fp, const bam_index_t *idx, int tid, int beg, int end, void *data, bam_fetch_f func);
595 @abstract Parse a region in the format: "chr2:100,000-200,000".
596 @discussion bam_header_t::hash will be initialized if empty.
597 @param header pointer to the header structure
598 @param str string to be parsed
599 @param ref_id the returned chromosome ID
600 @param begin the returned start coordinate
601 @param end the returned end coordinate
602 @return 0 on success; -1 on failure
604 int 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);