Do I need to sort and index a bam file?

[ejfuchs]

Hi all,

I have a large number of bam files from multiple individuals. I need to call snps on them using mpileup. When I sort and index with samtools, the I/O goes crazy on my cluster. Is the sorting/indexing necessary or does it just speed things up?

Thanks,

[krobison]

Almost certainly necessary. One way to think about it is you'll either pay this penalty now or pay it later -- any caller needs all the reads from the same region, and it is quick & convenient if they are together. I think most callers won't even call on unsorted data.

One thing to look at: are you sorting/indexing on an NFS drive & can you do the sort/index on a local drive to the machine. That might improve performance.

[ejfuchs]

Thanks for your reply. Yes, I'm indexing on the NFS, I'll try to code the bash to sort and index on the local drive and then move to central storage.

[swbarnes2]

Likely you will need to sort no matter what software you are using. For samtools mpileup, you actually don't need the .bam index, but those files are small, and fast to make, so you might as well make them.

[KevinLam]

one good reason you wanna sort/index your bam
is so that you can split your bams into chr or even regions so that you can parallelize your mpileup calls.