merging bwa alignments

[arnkas]

Hello,
I am using bwa to map solid data to the mouse genome, but get a segmentation fault while running bwa aln. If I separate the genome into 2 halves, it works all right, so I assume that the segfault was a memory problem. My question is: Is there a way to merge the result of mapping to 2 halves of the genome?

I have a similar question about dividing the reads into smaller subsets, while using the full database.

I appreciate any help on this.

[Lee Sam]

I've been doing BWA alignments and merging them using picard-tools "MergeSamFiles.jar"

[arnkas]

merging files which map reads to separate parts of the genome

Thank you for you response. I have also used samtools merge.

As I understand it, bwa eliminates reads that map to more than one place in the genome. Since my 2 files involves mapping reads to separate halves of the genome, reads that are mapped in both files should be eliminated. Does the merge process recognize this?

[nilshomer]

No it does not.

[lh3]

One should never split the genome. Split the read set. It is true for essentially all NGS aligners.

[arnkas]

Well, the reason I split the genome was that I ran into a segmentation fault running "bwa aln -c" against the entire mouse genome. I was able to isolate this to a single read. I tried it against various genomes, and it worked all right for smaller genomes, eg human mrna and half the mouse genome.

Here is the problem read:
@559_9_257_F3
GNANNNGGGGGNNNNNCNNNAGCACGAGAANNNNCGNGNGNTNCNNCNN
+
2!'!!!)*'&)!!!!!&!!!')&&.*)&+'!!!!(,!&!)!'!&!!&!!

I also assumed that there must be a way to postprocess a merged file from a split genome, although I haven't done it yet.