How deep to sequence?

[Mali Salmon]

Hi All
A friend of mine would like to find a specific virus from sludge samples, but she has no idea how deep she should sequence in order to actually see viral reads. The proportion between the virus and bacterial 16S is around 1:1000
Any ideas?
Thanks
Mali

[krobison]

I presume she is sequencing whole DNA without any prefiltering? Are there other DNAs present?

If the sample can be assumed to just be virus & bacteria, and the virus (bacteriophage?) genome is much, much smaller than the average bacterial genome size, then the proportion of viral DNA in the sample should be roughly:

(1 * average viral genome size ) / (1000 * average bacterial genome size)

It should then be possible to use standard sampling theory to estimate the probability of seeing at least n reads from the virus if you generate N reads. Given the genome size, one can then estimate the number of reads required to give a given coverage of the virus.