bwa not producing any alignments

[genec]

I'm trying to use bwa for the first time, and for some reason, even though it runs it's not actually producing any alignments.

This is what I'm doing:

bwa index -p hg19 -a bwtsw hg19.fasta

INPUT1=reads.1.fastq
INPUT2=reads.2.fastq
OUTPUT1=bwa_out.1.sai
OUTPUT2=bwa_out.2.sai
OUTPUT3=bwa_out.sam

bwa aln hg19 $INPUT1 > $OUTPUT1
bwa aln hg19 $INPUT2 > $OUTPUT2
bwa sampe hg19 $OUTPUT1 $OUTPUT2 $INPUT1 $INPUT2 > $OUTPUT3
ln -s bwa.sam A.sam

The input files are paired 76 nucleotide reads. The output consists of all the reads, unmapped:

READ_A 77      *       0       0       *       *       0       0       <sequence not shown>    <quality not shown>
READ_A 141     *       0       0       *       *       0       0       <sequence not shown>    <quality not shown>
...

Does anyone have any idea of what I'm missing here?

Thanks


** EDIT **
I forgot to mention that I'm running bwa-0.6.1 on Mac OS 10.6.8

[crf2s]

same problem

Hi genec,

Have you been able to correct your problem? I'm getting the exact same results!!! My sam file is identical to yours.

[genec]

I ended up running stampy which uses bwa as part of its pipeline -- that works fine. I never went back to spend more time on why bwa wasn't working properly when I tried to use it stand-alone.

Gene

[sdriscoll]

weird. i use bwa to align reads for SNP detection and i haven't had this issue. i'm running an older version though - 0.5.9-r16

[crf2s]

Hi genec,

So I thought it was odd that you and I were having the same issue and both running bwa 0.6.1 on Mac OS 10.6.8. So, I tried bwa 0.5.9 and it worked fine. The problem in the in bwa aln step in 0.6.1, but not sure why.

[mdimon]

I've had this problem also. I messed with bwa 0.6.1 for several hours. I did the indexing then aln step then sampe step in a variety of different ways on a couple different references. Each time bwa would act as if everything was fine (no error messages) but the output SAM file showed none of the reads aligned. Strangely, when I downloaded just chr8 from UCSC (as a shorter test case) and repeated the steps exactly, that DID work. Finally I downloaded bwa 0.5.9 and repeated my steps exactly and it's working fine now.

A few details:
-- I'm working on Mac OS X
-- I was testing 100,000 reads against the human genome (I tried the 1000Genomes version and the UCSC version of hg19)
-- one step that seems to be failing is the "[bwa_aln_core] calculate SA coordinate..." step. For my test case of 100,00 reads, the step took 0.5 seconds when bwa 0.6.1 failed, it took 16 seconds when bwa 0.6.1 worked against chr8 and it took 40 seconds for bwa 0.5.9 to align all 100,000 reads to the whole human genome.
-- the other difference I noticed is that my indexes seemed to be missing files when I ran the index in bwa 0.6.1. I was getting amb, ann, bwt, pac, and sa files (of the correct sizes) but not rbwt, rpac, or rsa files. To be fair, the chr8 test that worked with bwa 0.6.1 also didn't produce those files so maybe they aren't required anymore? Or maybe my chr8 test was a fluke.

I can provide more details if it will help anyone.

[sdriscoll]

so basically I shouldn't upgrade. just the other day i was pumped to find a new version. guess I'll wait a while...