1bwa(1) Bioinformatics tools bwa(1)
2
6 bwa - Burrows-Wheeler Alignment Tool
7
9 bwa index ref.fa
10 bwa mem ref.fa reads.fq > aln-se.sam
12 bwa mem ref.fa read1.fq read2.fq > aln-pe.sam
14 bwa aln ref.fa short_read.fq > aln_sa.sai
16 bwa samse ref.fa aln_sa.sai short_read.fq > aln-se.sam
18 bwa sampe ref.fa aln_sa1.sai aln_sa2.sai read1.fq read2.fq > aln-pe.sam
20 bwa bwasw ref.fa long_read.fq > aln.sam
22
25 BWA is a software package for mapping low-divergent sequences against a
26 large reference genome, such as the human genome. It consists of three
27 algorithms: BWA-backtrack, BWA-SW and BWA-MEM. The first algorithm is
28 designed for Illumina sequence reads up to 100bp, while the rest two
29 for longer sequences ranged from 70bp to 1Mbp. BWA-MEM and BWA-SW share
30 similar features such as long-read support and split alignment, but
31 BWA-MEM, which is the latest, is generally recommended for high-quality
32 queries as it is faster and more accurate. BWA-MEM also has better
33 performance than BWA-backtrack for 70-100bp Illumina reads.
34 For all the algorithms, BWA first needs to construct the FM-index for
36 the reference genome (the index command). Alignment algorithms are
37 invoked with different sub-commands: aln/samse/sampe for BWA-backtrack,
38 bwasw for BWA-SW and mem for the BWA-MEM algorithm.
39
42 index bwa index [-p prefix] [-a algoType] db.fa
43 Index database sequences in the FASTA format.
45 OPTIONS:
47 -p STR Prefix of the output database [same as db filename]
49 -a STR Algorithm for constructing BWT index. BWA implements
51 three algorithms for BWT construction: is, bwtsw and
52 rb2. The first algorithm is a little faster for small
53 database but requires large RAM and does not work for
54 databases with total length longer than 2GB. The sec‐
55 ond algorithm is adapted from the BWT-SW source code.
56 It in theory works with database with trillions of
57 bases. When this option is not specified, the appro‐
58 priate algorithm will be chosen automatically.
59 mem bwa mem [-aCHjMpP] [-t nThreads] [-k minSeedLen] [-w bandWidth]
62 [-d zDropoff] [-r seedSplitRatio] [-c maxOcc] [-D chainShadow]
63 [-m maxMateSW] [-W minSeedMatch] [-A matchScore] [-B mmPenalty]
64 [-O gapOpenPen] [-E gapExtPen] [-L clipPen] [-U unpairPen] [-x
65 readType] [-R RGline] [-H HDlines] [-v verboseLevel] db.prefix
66 reads.fq [mates.fq]
67 Align 70bp-1Mbp query sequences with the BWA-MEM algorithm.
69 Briefly, the algorithm works by seeding alignments with maximal
70 exact matches (MEMs) and then extending seeds with the affine-
71 gap Smith-Waterman algorithm (SW).
72 If mates.fq file is absent and option -p is not set, this com‐
74 mand regards input reads are single-end. If mates.fq is present,
75 this command assumes the i-th read in reads.fq and the i-th read
76 in mates.fq constitute a read pair. If -p is used, the command
77 assumes the 2i-th and the (2i+1)-th read in reads.fq constitute
78 a read pair (such input file is said to be interleaved). In this
79 case, mates.fq is ignored. In the paired-end mode, the mem com‐
80 mand will infer the read orientation and the insert size distri‐
81 bution from a batch of reads.
82 The BWA-MEM algorithm performs local alignment. It may produce
84 multiple primary alignments for different part of a query
85 sequence. This is a crucial feature for long sequences. However,
86 some tools such as Picard's markDuplicates does not work with
87 split alignments. One may consider to use option -M to flag
88 shorter split hits as secondary.
89 ALGORITHM OPTIONS:
92 -t INT Number of threads [1]
94 -k INT Minimum seed length. Matches shorter than INT will be
96 missed. The alignment speed is usually insensitive to
97 this value unless it significantly deviates from 20.
98 [19]
99 -w INT Band width. Essentially, gaps longer than INT will not
101 be found. Note that the maximum gap length is also
102 affected by the scoring matrix and the hit length, not
103 solely determined by this option. [100]
104 -d INT Off-diagonal X-dropoff (Z-dropoff). Stop extension
106 when the difference between the best and the current
107 extension score is above |i-j|*A+INT, where i and j
108 are the current positions of the query and reference,
109 respectively, and A is the matching score. Z-dropoff
110 is similar to BLAST's X-dropoff except that it doesn't
111 penalize gaps in one of the sequences in the align‐
112 ment. Z-dropoff not only avoids unnecessary extension,
113 but also reduces poor alignments inside a long good
114 alignment. [100]
115 -r FLOAT Trigger re-seeding for a MEM longer than min‐
117 SeedLen*FLOAT. This is a key heuristic parameter for
118 tuning the performance. Larger value yields fewer
119 seeds, which leads to faster alignment speed but lower
120 accuracy. [1.5]
121 -c INT Discard a MEM if it has more than INT occurence in the
123 genome. This is an insensitive parameter. [500]
124 -D FLOAT Drop chains shorter than FLOAT fraction of the longest
126 overlapping chain [0.5]
127 -m INT Perform at most INT rounds of mate-SW [50]
129 -W INT Drop a chain if the number of bases in seeds is
131 smaller than INT. This option is primarily used for
132 longer contigs/reads. When positive, it also affects
133 seed filtering. [0]
134 -P In the paired-end mode, perform SW to rescue missing
136 hits only but do not try to find hits that fit a
137 proper pair.
138 SCORING OPTIONS:
141 -A INT Matching score. [1]
143 -B INT Mismatch penalty. The sequence error rate is approxi‐
145 mately: {.75 * exp[-log(4) * B/A]}. [4]
146 -O INT[,INT]
148 Gap open penalty. If two numbers are specified, the
149 first is the penalty of openning a deletion and the
150 second for openning an insertion. [6]
151 -E INT[,INT]
153 Gap extension penalty. If two numbers are specified,
154 the first is the penalty of extending a deletion and
155 second for extending an insertion. A gap of length k
156 costs O + k*E (i.e. -O is for opening a zero-length
157 gap). [1]
158 -L INT[,INT]
160 Clipping penalty. When performing SW extension, BWA-
161 MEM keeps track of the best score reaching the end of
162 query. If this score is larger than the best SW score
163 minus the clipping penalty, clipping will not be
164 applied. Note that in this case, the SAM AS tag
165 reports the best SW score; clipping penalty is not
166 deduced. If two numbers are provided, the first is for
167 5'-end clipping and second for 3'-end clipping. [5]
168 -U INT Penalty for an unpaired read pair. BWA-MEM scores an
170 unpaired read pair as scoreRead1+scoreRead2-INT and
171 scores a paired as scoreRead1+scoreRead2-insert‐
172 Penalty. It compares these two scores to determine
173 whether we should force pairing. A larger value leads
174 to more aggressive read pair. [17]
175 -x STR Read type. Changes multiple parameters unless overri‐
177 den [null]
178 pacbio: -k17 -W40 -r10 -A1 -B1 -O1 -E1 -L0 (PacBio
180 reads to ref)
181 ont2d: -k14 -W20 -r10 -A1 -B1 -O1 -E1 -L0 (Oxford
183 Nanopore 2D-reads to ref)
184 intractg: -B9 -O16 -L5 (intra-species contigs to ref)
186 INPUT/OUTPUT OPTIONS:
188 -p Smart pairing. If two adjacent reads have the same
190 name, they are considered to form a read pair. This
191 way, paired-end and single-end reads can be mixed in a
192 single FASTA/Q stream.
193 -R STR Complete read group header line. '\t' can be used in
195 STR and will be converted to a TAB in the output SAM.
196 The read group ID will be attached to every read in
197 the output. An example is '@RG\tID:foo\tSM:bar'.
198 [null]
199 -H ARG If ARG starts with @, it is interpreted as a string
201 and gets inserted into the output SAM header; other‐
202 wise, ARG is interpreted as a file with all lines
203 starting with @ in the file inserted into the SAM
204 header. [null]
205 -o FILE Write the output SAM file to FILE. For compatibility
207 with other BWA commands, this option may also be given
208 as -f FILE. [standard ouptut]
209 -q
211 Don't reduce the mapping quality of split alignment
212 of lower alignment score.
213 -5 For split alignment, mark the segment with the small‐
215 est coordinate as the primary. It automatically
216 applies option -q as well. This option may help some
217 Hi-C pipelines. By default, BWA-MEM marks highest
218 scoring segment as primary.
219 -K INT Process INT input bases in each batch regardless of
221 the number of threads in use [10000000*nThreads]. By
222 default, the batch size is proportional to the number
223 of threads in use. Because the inferred insert size
224 distribution slightly depends on the batch size, using
225 different number of threads may produce different out‐
226 put. Specifying this option helps reproducibility.
227 -T INT Don't output alignment with score lower than INT.
229 This option affects output and occasionally SAM flag
230 2. [30]
231 -j Treat ALT contigs as part of the primary assembly
233 (i.e. ignore the db.prefix.alt file).
234 -h INT[,INT2]
236 If a query has not more than INT hits with score
237 higher than 80% of the best hit, output them all in
238 the XA tag. If INT2 is specified, BWA-MEM outputs up
239 to INT2 hits if the list contains a hit to an ALT con‐
240 tig. [5,200]
241 -a Output all found alignments for single-end or unpaired
243 paired-end reads. These alignments will be flagged as
244 secondary alignments.
245 -C Append FASTA/Q comment to SAM output. This option can
247 be used to transfer read meta information (e.g. bar‐
248 code) to the SAM output. Note that the FASTA/Q comment
249 (the string after a space in the header line) must
250 conform the SAM spec (e.g. BC:Z:CGTAC). Malformated
251 comments lead to incorrect SAM output.
252 -Y Use soft clipping CIGAR operation for supplementary
254 alignments. By default, BWA-MEM uses soft clipping for
255 the primary alignment and hard clipping for supplemen‐
256 tary alignments.
257 -M Mark shorter split hits as secondary (for Picard com‐
259 patibility).
260 -v INT Control the verbosity level of the output. This option
262 has not been fully supported throughout BWA. Ideally,
263 a value 0 for disabling all the output to stderr; 1
264 for outputting errors only; 2 for warnings and errors;
265 3 for all normal messages; 4 or higher for debugging.
266 When this option takes value 4, the output is not SAM.
267 [3]
268 -I FLOAT[,FLOAT[,INT[,INT]]]
270 Specify the mean, standard deviation (10% of the mean
271 if absent), max (4 sigma from the mean if absent) and
272 min (4 sigma if absent) of the insert size distribu‐
273 tion. Only applicable to the FR orientation. By
274 default, BWA-MEM infers these numbers and the pair
275 orientations given enough reads. [inferred]
276 aln bwa aln [-n maxDiff] [-o maxGapO] [-e maxGapE] [-d nDelTail] [-i
280 nIndelEnd] [-k maxSeedDiff] [-l seedLen] [-t nThrds] [-cRN] [-M
281 misMsc] [-O gapOsc] [-E gapEsc] [-q trimQual] <in.db.fasta>
282 <in.query.fq> > <out.sai>
283 Find the SA coordinates of the input reads. Maximum maxSeedDiff
285 differences are allowed in the first seedLen subsequence and
286 maximum maxDiff differences are allowed in the whole sequence.
287 OPTIONS:
289 -n NUM Maximum edit distance if the value is INT, or the
291 fraction of missing alignments given 2% uniform base
292 error rate if FLOAT. In the latter case, the maximum
293 edit distance is automatically chosen for different
294 read lengths. [0.04]
295 -o INT Maximum number of gap opens [1]
297 -e INT Maximum number of gap extensions, -1 for k-difference
299 mode (disallowing long gaps) [-1]
300 -d INT Disallow a long deletion within INT bp towards the
302 3'-end [16]
303 -i INT Disallow an indel within INT bp towards the ends [5]
305 -l INT Take the first INT subsequence as seed. If INT is
307 larger than the query sequence, seeding will be dis‐
308 abled. For long reads, this option is typically ranged
309 from 25 to 35 for `-k 2'. [inf]
310 -k INT Maximum edit distance in the seed [2]
312 -t INT Number of threads (multi-threading mode) [1]
314 -M INT Mismatch penalty. BWA will not search for suboptimal
316 hits with a score lower than (bestScore-misMsc). [3]
317 -O INT Gap open penalty [11]
319 -E INT Gap extension penalty [4]
321 -R INT Proceed with suboptimal alignments if there are no
323 more than INT equally best hits. This option only
324 affects paired-end mapping. Increasing this threshold
325 helps to improve the pairing accuracy at the cost of
326 speed, especially for short reads (~32bp).
327 -c Reverse query but not complement it, which is required
329 for alignment in the color space. (Disabled since
330 0.6.x)
331 -N Disable iterative search. All hits with no more than
333 maxDiff differences will be found. This mode is much
334 slower than the default.
335 -q INT Parameter for read trimming. BWA trims a read down to
337 argmax_x{\sum_{i=x+1}^l(INT-q_i)} if q_l<INT where l
338 is the original read length. [0]
339 -I The input is in the Illumina 1.3+ read format (quality
341 equals ASCII-64).
342 -B INT Length of barcode starting from the 5'-end. When INT
344 is positive, the barcode of each read will be trimmed
345 before mapping and will be written at the BC SAM tag.
346 For paired-end reads, the barcode from both ends are
347 concatenated. [0]
348 -b Specify the input read sequence file is the BAM for‐
350 mat. For paired-end data, two ends in a pair must be
351 grouped together and options -1 or -2 are usually
352 applied to specify which end should be mapped. Typical
353 command lines for mapping pair-end data in the BAM
354 format are:
355 bwa aln ref.fa -b1 reads.bam > 1.sai
357 bwa aln ref.fa -b2 reads.bam > 2.sai
358 bwa sampe ref.fa 1.sai 2.sai reads.bam reads.bam >
359 aln.sam
360 -0 When -b is specified, only use single-end reads in
362 mapping.
363 -1 When -b is specified, only use the first read in a
365 read pair in mapping (skip single-end reads and the
366 second reads).
367 -2 When -b is specified, only use the second read in a
369 read pair in mapping.
370 samse bwa samse [-n maxOcc] <in.db.fasta> <in.sai> <in.fq> > <out.sam>
373 Generate alignments in the SAM format given single-end reads.
375 Repetitive hits will be randomly chosen.
376 OPTIONS:
378 -n INT Maximum number of alignments to output in the XA tag
380 for reads paired properly. If a read has more than INT
381 hits, the XA tag will not be written. [3]
382 -r STR Specify the read group in a format like
384 `@RG\tID:foo\tSM:bar'. [null]
385 sampe bwa sampe [-a maxInsSize] [-o maxOcc] [-n maxHitPaired] [-N max‐
388 HitDis] [-P] <in.db.fasta> <in1.sai> <in2.sai> <in1.fq> <in2.fq>
389 > <out.sam>
390 Generate alignments in the SAM format given paired-end reads.
392 Repetitive read pairs will be placed randomly.
393 OPTIONS:
395 -a INT Maximum insert size for a read pair to be considered
397 being mapped properly. Since 0.4.5, this option is only
398 used when there are not enough good alignment to infer
399 the distribution of insert sizes. [500]
400 -o INT Maximum occurrences of a read for pairing. A read with
402 more occurrneces will be treated as a single-end read.
403 Reducing this parameter helps faster pairing. [100000]
404 -P Load the entire FM-index into memory to reduce disk
406 operations (base-space reads only). With this option, at
407 least 1.25N bytes of memory are required, where N is the
408 length of the genome.
409 -n INT Maximum number of alignments to output in the XA tag for
411 reads paired properly. If a read has more than INT hits,
412 the XA tag will not be written. [3]
413 -N INT Maximum number of alignments to output in the XA tag for
415 disconcordant read pairs (excluding singletons). If a
416 read has more than INT hits, the XA tag will not be
417 written. [10]
418 -r STR Specify the read group in a format like
420 `@RG\tID:foo\tSM:bar'. [null]
421 bwasw bwa bwasw [-a matchScore] [-b mmPen] [-q gapOpenPen] [-r
424 gapExtPen] [-t nThreads] [-w bandWidth] [-T thres] [-s hspIntv]
425 [-z zBest] [-N nHspRev] [-c thresCoef] <in.db.fasta> <in.fq>
426 [mate.fq]
427 Align query sequences in the in.fq file. When mate.fq is
429 present, perform paired-end alignment. The paired-end mode only
430 works for reads Illumina short-insert libraries. In the paired-
431 end mode, BWA-SW may still output split alignments but they are
432 all marked as not properly paired; the mate positions will not
433 be written if the mate has multiple local hits.
434 OPTIONS:
436 -a INT Score of a match [1]
438 -b INT Mismatch penalty [3]
440 -q INT Gap open penalty [5]
442 -r INT Gap extension penalty. The penalty for a contiguous
444 gap of size k is q+k*r. [2]
445 -t INT Number of threads in the multi-threading mode [1]
447 -w INT Band width in the banded alignment [33]
449 -T INT Minimum score threshold divided by a [37]
451 -c FLOAT Coefficient for threshold adjustment according to
453 query length. Given an l-long query, the threshold for
454 a hit to be retained is a*max{T,c*log(l)}. [5.5]
455 -z INT Z-best heuristics. Higher -z increases accuracy at the
457 cost of speed. [1]
458 -s INT Maximum SA interval size for initiating a seed. Higher
460 -s increases accuracy at the cost of speed. [3]
461 -N INT Minimum number of seeds supporting the resultant
463 alignment to skip reverse alignment. [5]
464
467 The output of the `aln' command is binary and designed for BWA use
468 only. BWA outputs the final alignment in the SAM (Sequence Align‐
469 ment/Map) format. Each line consists of:
470 ┌────┬───────┬──────────────────────────────────────────────────────────┐
473 │Col │ Field │ Description │
474 ├────┼───────┼──────────────────────────────────────────────────────────┤
475 │ 1 │ QNAME │ Query (pair) NAME │
476 │ 2 │ FLAG │ bitwise FLAG │
477 │ 3 │ RNAME │ Reference sequence NAME │
478 │ 4 │ POS │ 1-based leftmost POSition/coordinate of clipped sequence │
479 │ 5 │ MAPQ │ MAPping Quality (Phred-scaled) │
480 │ 6 │ CIAGR │ extended CIGAR string │
481 │ 7 │ MRNM │ Mate Reference sequence NaMe (`=' if same as RNAME) │
482 │ 8 │ MPOS │ 1-based Mate POSistion │
483 │ 9 │ ISIZE │ Inferred insert SIZE │
484 │10 │ SEQ │ query SEQuence on the same strand as the reference │
485 │11 │ QUAL │ query QUALity (ASCII-33 gives the Phred base quality) │
486 │12 │ OPT │ variable OPTional fields in the format TAG:VTYPE:VALUE │
487 └────┴───────┴──────────────────────────────────────────────────────────┘
488 Each bit in the FLAG field is defined as:
490 ┌────┬────────┬───────────────────────────────────────┐
493 │Chr │ Flag │ Description │
494 ├────┼────────┼───────────────────────────────────────┤
495 │ p │ 0x0001 │ the read is paired in sequencing │
496 │ P │ 0x0002 │ the read is mapped in a proper pair │
497 │ u │ 0x0004 │ the query sequence itself is unmapped │
498 │ U │ 0x0008 │ the mate is unmapped │
499 │ r │ 0x0010 │ strand of the query (1 for reverse) │
500 │ R │ 0x0020 │ strand of the mate │
501 │ 1 │ 0x0040 │ the read is the first read in a pair │
502 │ 2 │ 0x0080 │ the read is the second read in a pair │
503 │ s │ 0x0100 │ the alignment is not primary │
504 │ f │ 0x0200 │ QC failure │
505 │ d │ 0x0400 │ optical or PCR duplicate │
506 │ S │ 0x0800 │ supplementary alignment │
507 └────┴────────┴───────────────────────────────────────┘
508 The Please check <http://samtools.sourceforge.net> for the format spec‐
510 ification and the tools for post-processing the alignment.
511 BWA generates the following optional fields. Tags starting with `X' are
513 specific to BWA.
514 ┌────┬──────────────────────────────────────────────────┐
517 │Tag │ Meaning │
518 ├────┼──────────────────────────────────────────────────┤
519 │NM │ Edit distance │
520 │MD │ Mismatching positions/bases │
521 │AS │ Alignment score │
522 │BC │ Barcode sequence │
523 │SA │ Supplementary alignments │
524 ├────┼──────────────────────────────────────────────────┤
525 │X0 │ Number of best hits │
526 │X1 │ Number of suboptimal hits found by BWA │
527 │XN │ Number of ambiguous bases in the referenece │
528 │XM │ Number of mismatches in the alignment │
529 │XO │ Number of gap opens │
530 │XG │ Number of gap extentions │
531 │XT │ Type: Unique/Repeat/N/Mate-sw │
532 │XA │ Alternative hits; format: /(chr,pos,CIGAR,NM;)*/ │
533 ├────┼──────────────────────────────────────────────────┤
534 │XS │ Suboptimal alignment score │
535 │XF │ Support from forward/reverse alignment │
536 │XE │ Number of supporting seeds │
537 └────┴──────────────────────────────────────────────────┘
538 Note that XO and XG are generated by BWT search while the CIGAR string
540 by Smith-Waterman alignment. These two tags may be inconsistent with
541 the CIGAR string. This is not a bug.
542
545 Alignment Accuracy
546 When seeding is disabled, BWA guarantees to find an alignment contain‐
547 ing maximum maxDiff differences including maxGapO gap opens which do
548 not occur within nIndelEnd bp towards either end of the query. Longer
549 gaps may be found if maxGapE is positive, but it is not guaranteed to
550 find all hits. When seeding is enabled, BWA further requires that the
551 first seedLen subsequence contains no more than maxSeedDiff differ‐
552 ences.
553 When gapped alignment is disabled, BWA is expected to generate the same
555 alignment as Eland version 1, the Illumina alignment program. However,
556 as BWA change `N' in the database sequence to random nucleotides, hits
557 to these random sequences will also be counted. As a consequence, BWA
558 may mark a unique hit as a repeat, if the random sequences happen to be
559 identical to the sequences which should be unqiue in the database.
560 By default, if the best hit is not highly repetitive (controlled by
562 -R), BWA also finds all hits contains one more mismatch; otherwise, BWA
563 finds all equally best hits only. Base quality is NOT considered in
564 evaluating hits. In the paired-end mode, BWA pairs all hits it found.
565 It further performs Smith-Waterman alignment for unmapped reads to res‐
566 cue reads with a high erro rate, and for high-quality anomalous pairs
567 to fix potential alignment errors.
568 Estimating Insert Size Distribution
571 BWA estimates the insert size distribution per 256*1024 read pairs. It
572 first collects pairs of reads with both ends mapped with a single-end
573 quality 20 or higher and then calculates median (Q2), lower and higher
574 quartile (Q1 and Q3). It estimates the mean and the variance of the
575 insert size distribution from pairs whose insert sizes are within
576 interval [Q1-2(Q3-Q1), Q3+2(Q3-Q1)]. The maximum distance x for a pair
577 considered to be properly paired (SAM flag 0x2) is calculated by solv‐
578 ing equation Phi((x-mu)/sigma)=x/L*p0, where mu is the mean, sigma is
579 the standard error of the insert size distribution, L is the length of
580 the genome, p0 is prior of anomalous pair and Phi() is the standard
581 cumulative distribution function. For mapping Illumina short-insert
582 reads to the human genome, x is about 6-7 sigma away from the mean.
583 Quartiles, mean, variance and x will be printed to the standard error
584 output.
585 Memory Requirement
588 With bwtsw algorithm, 5GB memory is required for indexing the complete
589 human genome sequences. For short reads, the aln command uses ~3.2GB
590 memory and the sampe command uses ~5.4GB.
591 Speed
594 Indexing the human genome sequences takes 3 hours with bwtsw algorithm.
595 Indexing smaller genomes with IS algorithms is faster, but requires
596 more memory.
597 The speed of alignment is largely determined by the error rate of the
599 query sequences (r). Firstly, BWA runs much faster for near perfect
600 hits than for hits with many differences, and it stops searching for a
601 hit with l+2 differences if a l-difference hit is found. This means BWA
602 will be very slow if r is high because in this case BWA has to visit
603 hits with many differences and looking for these hits is expensive.
604 Secondly, the alignment algorithm behind makes the speed sensitive to
605 [k log(N)/m], where k is the maximum allowed differences, N the size of
606 database and m the length of a query. In practice, we choose k w.r.t. r
607 and therefore r is the leading factor. I would not recommend to use BWA
608 on data with r>0.02.
609 Pairing is slower for shorter reads. This is mainly because shorter
611 reads have more spurious hits and converting SA coordinates to chromo‐
612 somal coordinates are very costly.
613
616 Since version 0.6, BWA has been able to work with a reference genome
617 longer than 4GB. This feature makes it possible to integrate the for‐
618 ward and reverse complemented genome in one FM-index, which speeds up
619 both BWA-short and BWA-SW. As a tradeoff, BWA uses more memory because
620 it has to keep all positions and ranks in 64-bit integers, twice larger
621 than 32-bit integers used in the previous versions.
622 The latest BWA-SW also works for paired-end reads longer than 100bp. In
624 comparison to BWA-short, BWA-SW tends to be more accurate for highly
625 unique reads and more robust to relative long INDELs and structural
626 variants. Nonetheless, BWA-short usually has higher power to distin‐
627 guish the optimal hit from many suboptimal hits. The choice of the map‐
628 ping algorithm may depend on the application.
629
632 BWA website <http://bio-bwa.sourceforge.net>, Samtools website
633 <http://samtools.sourceforge.net>
634
637 Heng Li at the Sanger Institute wrote the key source codes and inte‐
638 grated the following codes for BWT construction: bwtsw
639 <http://i.cs.hku.hk/~ckwong3/bwtsw/>, implemented by Chi-Kwong Wong at
640 the University of Hong Kong and IS
641 <http://yuta.256.googlepages.com/sais> originally proposed by Nong Ge
642 <http://www.cs.sysu.edu.cn/nong/> at the Sun Yat-Sen University and
643 implemented by Yuta Mori.
644
647 The full BWA package is distributed under GPLv3 as it uses source codes
648 from BWT-SW which is covered by GPL. Sorting, hash table, BWT and IS
649 libraries are distributed under the MIT license.
650 If you use the BWA-backtrack algorithm, please cite the following
652 paper:
653 Li H. and Durbin R. (2009) Fast and accurate short read alignment with
655 Burrows-Wheeler transform. Bioinformatics, 25, 1754-1760. [PMID:
656 19451168]
657 If you use the BWA-SW algorithm, please cite:
659 Li H. and Durbin R. (2010) Fast and accurate long-read alignment with
661 Burrows-Wheeler transform. Bioinformatics, 26, 589-595. [PMID:
662 20080505]
663 If you use BWA-MEM or the fastmap component of BWA, please cite:
665 Li H. (2013) Aligning sequence reads, clone sequences and assembly con‐
667 tigs with BWA-MEM. arXiv:1303.3997v1 [q-bio.GN].
668 It is likely that the BWA-MEM manuscript will not appear in a peer-
670 reviewed journal.
671
674 BWA is largely influenced by BWT-SW. It uses source codes from BWT-SW
675 and mimics its binary file formats; BWA-SW resembles BWT-SW in several
676 ways. The initial idea about BWT-based alignment also came from the
677 group who developed BWT-SW. At the same time, BWA is different enough
678 from BWT-SW. The short-read alignment algorithm bears no similarity to
679 Smith-Waterman algorithm any more. While BWA-SW learns from BWT-SW, it
680 introduces heuristics that can hardly be applied to the original algo‐
681 rithm. In all, BWA does not guarantee to find all local hits as what
682 BWT-SW is designed to do, but it is much faster than BWT-SW on both
683 short and long query sequences.
684 I started to write the first piece of codes on 24 May 2008 and got the
686 initial stable version on 02 June 2008. During this period, I was
687 acquainted that Professor Tak-Wah Lam, the first author of BWT-SW
688 paper, was collaborating with Beijing Genomics Institute on SOAP2, the
689 successor to SOAP (Short Oligonucleotide Analysis Package). SOAP2 has
690 come out in November 2008. According to the SourceForge download page,
691 the third BWT-based short read aligner, bowtie, was first released in
692 August 2008. At the time of writing this manual, at least three more
693 BWT-based short-read aligners are being implemented.
694 The BWA-SW algorithm is a new component of BWA. It was conceived in No‐
696 vember 2008 and implemented ten months later.
697 The BWA-MEM algorithm is based on an algorithm finding super-maximal
699 exact matches (SMEMs), which was first published with the fermi assem‐
700 bler paper in 2012. I first implemented the basic SMEM algorithm in the
701 fastmap command for an experiment and then extended the basic algorithm
702 and added the extension part in Feburary 2013 to make BWA-MEM a fully
703 featured mapper.
704bwa-0.7.17-r1188 23 October 2017 bwa(1)