I'm interested in sequencing a sputum sample using an ONT MinION without performing any type of DNA amplification. Has anyone found a good reference (or anecdotal evidence) for a range of human to pathogen DNA ratios typically found in sputum samples? Does this depend on whether the patient is infected by a virus or bacteria, and the bacterial species? Can I expect to find 10x as much human DNA as pathogen DNA, or 100x-1000x? I would really just like a ballpark estimate.
My recollection of a conversation I had with Justin O'Grady is that it's about 99% host DNA for unfiltered samples. There are a few ways to reduce that portion prior to sequencing; he has written a paper that mentions a saponin-based host depletion method that is quite effective: