[CCREPE] Is it okay to use input data whose row-sums are not 1 ? (due to filtering after normzliation)


I want to do network analysis for certain core-taxa (whose prevalences are more than threshold.)

For this, I first prepared TSS normalized species table, and then filtered out taxa based on prevalence cut-off. (Therefore, the row-sums are not 1 and different across samples.)

In the ccrepe manual, there is a saying that one should use compositional data (with constant row sum) for input data.
But, in my case, it is compositional data but the row sums are not constant because of the filtering process.

Can I use this matrix as ccrepe input?