I am trying to normalize a dataset that has a lot of zeros. Using the type="iterate" did not work for estimateSizeFactors, but adding 1 as a pseudocount to my ASV matrix did. However, I am wondering what I should do with the normalized counts after this... Do I just subtract out what the former-zero values are now (>0)? Or does it not make a difference to leave them in as is for downstream analyses?

Thanks, Devin

Maybe DESeq2 alone isn't a good model for this data, or you could use zinbwave in combination with DESeq2?