bwa sampe error ... bad database format?

[rwarren]

Hi,
I am running bwa 0.5.9 on a custom, indexed, sequence database comprised of various genbank sequences.

When I run:

bwa aln gb.fa x_1.fq > a.sai
bwa aln gb.fa x_2.fq > b.sai
bwa sampe gb.fa a.sai b.sai x_1.fq x_2.fq > gb.sam

bwa crashes on bwa sampe with:

[bns_restore_core] fail to open file ' VirtualSequence=""></div><div id="viewercontent13" class="seq gbff" val="381286149" SequenceSize="29462" VirtualSequence=""></div><div id="viewercontent14" class="seq gbff" val="381283770" SequenceSize="35021" VirtualSequence=""></div><div id="viewercontent15" class="seq gbff" val="381286161" SequenceSize="29364" VirtualSequence=""></div><div id="viewercontent16" class="seq gbff" val="381282985" SequenceSize="9499" VirtualSequence=""></div><div id="viewercontent17" class="seq gbff" val="381216220" SequenceSize="16093" VirtualSequence=""></div><div id="viewercontent18" class="seq gbff" val="381147283" SequenceSize="5929" VirtualSequence=""></div><div id="viewercontent19" class="seq gbff" val="380307325" SequenceSize="10631" VirtualSequence=""></div><div id="viewercontent20" class="seq gbff" val="380862749" SequenceSize="6588" VirtualSequence=""></div>'. Abort!
./test.sh: line 3: 25171 Aborted bwa sampe gb.fa a.sai b.sai x_1.fq x_2.fq > gb.sam

The 9GB gb.fa database file was indexed with default, as per the instructions provided and no errors were returned from the indexing.

When I run the same sets of bwa commands using a different, human genome database, there are no errors.

I was wondering whether characters others than ACGTN in the seq. db would make bwa sampe crash (there are such characters in the gb.fa)? Has anyone seen an error like this?

Thanks for your help
Rene

[xied75]

That block looks like XML? Is that really a FA file?

[rwarren]

it is, now reformatting to remove any offending characters.
The interesting thing about this error is that, usually, when bwa returns a fail to open file rror, the file is in quotation, not some XML block:

eg.
[bns_restore_core] fail to open file '/projects/02/rwarren/nt_db/2012/humanNCRNA.fa.fa.ann'. Abort!
./testdb.sh: line 3: 30215 Aborted /home/pubseq/BioSw/bwa/bwa-0.5.9/bwa sampe /projects/02/rwarren/nt_db/2012/humanNCRNA.fa.fa 1.sai 2.sai unmappedSorted.bam unmappedSorted.bam > test.sam

bizarre.

[rwarren]

i confirm that junk in the custom sequence database generated the error.
issue resolved.

[xied75]

I just did a test, I modified a .fasta with two illegal char 'R', 'S', and then run INDEX, ALN, and SAMPE, everything still works, although output SAM files are different. So BWA should be able to cope with illegal characters.

And if you open the .amb file, you should see the illegal chars listed there, in my case:

249240584 1 R
249240585 1 S

So if your fasta is big, and really messed with many illegal chars, then your .amb file should be very very big.

Besides, your error msg is BWA complaining a file can't be opened, at this stage, .bwt, .sa, and .pac are not opened yet, so having illegal chars shouldn't be the reason for this error.