I am working with digital droplet PCR data (http://www.bio-rad.com/webroot/web/pdf/lsr/literature/Bulletin_6407.pdf). Essentially, the read out of such experiments are the number of "positive" and "negative" droplets for a given probe of interest. This experiment can be applied for somatic mutation detection where a a probe can be designed a particular position and then detects how many droplets are positive for the mutation.
Ideally each droplet contains one DNA molecule, but this is not always the case. To correct for this, a poission distribution is used with a correction such that:
Copies per droplet = -ln(1-p) where p = fraction of positive droplets
This gives us a poisson corrected value (Figure 1.12 from link above) which from my understanding is equivalent to a variant ratio. If you run such an experiment for a tumour-normal pair, you could do a copies per droplet value for both the tumour and normal.
My question is what statistical test would you employ to test whether a position is somatic or not between tumour-normal when the data is poisson corrected? Is it a two-sample poisson test?