Is it convinient to discretize tpm-RNASeq expression matrix before building a bayesian network?

Question

[rodguinea]

Hi. I want to build a Bayesian network (using the MXM package https://rdrr.io/cran/MXM/) from RNASeq expression data (normalized using the TPM method) to find causal relations. What I want to know is if it is convenient to pre-process the normalized expression matrix using a discretization method (http://www.bnlearn.com/documentation/man/preprocessing.html).

*Is a cross species expression matrix (multiple species).

Thanks a lot, Rodrigo