[DavyK]

I have only recently discovered this problem.
Compiling using make runs ok, no errors.
Indexing (human_g1k_v37 from the GATK website) on mac takes around an hour, as standard (I think).
bwa aln runs with no errors, but as the OP mentions each step takes almost no time.
bwa sampe also runs with no errors but fails to infer insert size at every step. The resulting sam file has no alignment coordinates in it, and flagstat shows 0% mapped reads.
After transferring data to linux cluster, and rebuilding the index, everything seems ok and flagstat shows 89% of the reads have been mapped.

I wonder is there some kind of library that a Standard Mac OSX lion is missing. Although is that was the case surely bwa would throw an error when it fails to import that library or package. I'm totally ignorant of C though so I'm guessing here.

[GenoMax]

Is anyone prepared to host a working version of the BWA index for the human genome?
It would be about 4.5GB. I think it would take less than 2weeks to down load and I might
have more confidence in it than either of the two version I have created using
bwa index /tmp/INPUTS.fa -p h_sapiens_37.5 -a bwtsw

Many thanks
Bill

Bill,

Illumina built indexes are available at the iGenomes site.

http://cufflinks.cbcb.umd.edu/igenomes.html

[wlangdon]

It seems to me building the bwa indexes is error prone.
I am just trying it again today and despite running what is supposed to be an
identical script, I get different output (eg the first value in h_sapiens_37.5.amb is half
the size of the value in the version bwt produced two weeks ago).

Is anyone prepared to host a working version of the BWA index for the human genome?
It would be about 4.5GB. I think it would take less than 2weeks to down load and I might
have more confidence in it than either of the two version I have created using
bwa index /tmp/INPUTS.fa -p h_sapiens_37.5 -a bwtsw

Many thanks
Bill

[TabeaK]

I realise that, currently running some single end alignments, which take longer...

But anyway, the software does not produce weird faults as above, all my reads get processed and by "fast"; I still mean several hours.

[cboustred]

It 'worked' lightening fast for me too, except when I looked at the resulting BAM files none of the reads were aligned properly

Try running samtools flagstat to check the number of mapped reads

[TabeaK]

Uhm, by the way, I have started running the latest bwa on my Macbook (Mac OSX 10.7.4; Lion) yesterday.

Runs fine - the first round of paired end alignment is complete, the program ran right through and processed my 60million odd reads.

The only issues I encountered were with bwa having problems to infer the insert size, but I suspect that is due to either a problem with my fastq files or my ineptitude at selecting the right settings.

The program itself works and is lightning-fast even on my crappy machine!

Edit: I actually had problems with directing bwa to the file directories. No idea why, but even though the path to the files WAS DEFINITELY right, bwa failed to open the files. So, I have copied everything to my home directory and run from there --> no more problems.

[cboustred]

I am also having the same problem running the current bwa release (0.6.2) on Mac OSX Moutain Lion (10.8.2). All indexes were made on the Mac using bwa.

Using the older version of bwa (0.5.9) everything works fine

[quincymo]

Thanks for your replies.

My Mac OS version is Mac OS X Lion 10.7.5.

[GenoMax]

It failed for all the files I tested. But bwa 0.5.9 worked fine on the same files.

Perhaps someone else on the forum can chime in (do not have a Mac handy to test now) if they are successfully using bwa (v.0.6.2) on OS X. Please indicate exact OS X version.

[quincymo]

It failed for all the files I tested. But bwa 0.5.9 worked fine on the same files.

[GenoMax]

I think it looks OK.

2.9G genome.fa
8.0K genome.fa.amb
4.0K genome.fa.ann
2.9G genome.fa.bwt
738M genome.fa.pac
1.4G genome.fa.sa
8.0G total

That looks good.

Is the alignment failing for all files or just one file?

[quincymo]

bwa 0.5.9 works fine on my Mac, but 0.6.x do not work.

[quincymo]

I think it looks OK.

2.9G genome.fa
8.0K genome.fa.amb
4.0K genome.fa.ann
2.9G genome.fa.bwt
738M genome.fa.pac
1.4G genome.fa.sa
8.0G total

[GenoMax]

I did build the index on Mac
bwa index -a bwtsw genome.fa
which produced the following files:
genome.fa.ann genome.fa.pac genome.fa.amb genome.fa.bwt genome.fa.sa

File sizes for the index files look ok?
e.g. .sa file should be about 1.5G and .bwt 2.9G.

[quincymo]

I did build the index on Mac
bwa index -a bwtsw genome.fa
which produced the following files:
genome.fa.ann genome.fa.pac genome.fa.amb genome.fa.bwt genome.fa.sa

bwa failed to calculate SA coordinate. Note it only took 0.75 sec.

Software/bwa-0.6.2/bwa aln Genome/Homo_sapiens/UCSC/hg19/Sequence/BWAIndex/version0.6.2/genome.fa ERR001014_1.fastq > ERR001014_1.sai
[bwa_aln] 17bp reads: max_diff = 2
[bwa_aln] 38bp reads: max_diff = 3
[bwa_aln] 64bp reads: max_diff = 4
[bwa_aln] 93bp reads: max_diff = 5
[bwa_aln] 124bp reads: max_diff = 6
[bwa_aln] 157bp reads: max_diff = 7
[bwa_aln] 190bp reads: max_diff = 8
[bwa_aln] 225bp reads: max_diff = 9
[bwa_aln_core] calculate SA coordinate... 0.75 sec
[bwa_aln_core] write to the disk... 0.02 sec
[bwa_aln_core] 250000 sequences have been processed.
[main] Version: 0.6.2-r126
[main] CMD:Software/bwa-0.6.2/bwa aln Genome/Homo_sapiens/UCSC/hg19/Sequence/BWAIndex/version0.6.2/genome.fa ERR001014_1.fastq
[main] Real time: 1.118 sec; CPU: 1.114 sec