Thanks for your attention!!
Referring to my previous question: (Alternative way to compare the KOs from PICRUSt2 result) and suggestions from @jwdebelius,
I compare my KOs generated from PICRUSt2 by DEseq2. However, I realize that KO counts from PICRUSt2 are not integers, which generates error while running DEseq2.
I would like to know why PICRUSt2 generates non-integers KO counts? Could I just round up the counts and go for the downstream analysis?