How are the read counts normalized to estimate ploidy?

Briefly, reads for every amplicon are modeled using the spike-in cell line. The reads to every amplicon are modeled using a negative binomial distribution with the total read depth of the cells as an extra parameter to model differences in droplet PCR efficiency. Using this model the likelihood for various copy numbers are estimated for each cell and amplicon on which the prediction has to be made. The copy number which is the most likely is predicted to be the ploidy for that cell and amplicon.