Coverage required for SNP detection?

[Kasycas]

Hi all,

I am sequencing bacterial genomes, certain genomic strains of which have been published but not the exact strains I'll be trying to detect SNPs for.

First off, as my title asks, can anyone with experience (or opinions) of SNP detection tell me what coverage you need to reliably detect SNPs??

I am also looking at the best possible technology for the sequencing. 454 is out because of homopolymers. I have experience analysing Illumina data so I was thinking of going with paired end 100bps, however one of the sequencing companies suggested SOLiD. I've never used SOLiD data before and hence, I'm reluctant but if it's good for the purpose then I should just bite the bullet...

Any recommendations?

Thanks!

[NextGenSeq]

Most people aim for at least 20X coverage for SNP and indel detection.

[xgenseq]

-- Minimum 20X coverage
-- when predicting putative SNP use minimum 10 reads with occurrence of each variant 4x, give reasonably good and nearly accurate SNP.

[Chipper]

20x is for diploid genomes. But I guess for bacterial genomes it is difficult to get less than 20x coverage anyway even if you multiplex samples (at least for SOLiD and HiSeq).