Zero imputation prior adjust_batch application

Downstream analysis and statistics
1

Hello, I’m studying 16S data from 20 CRC mice in R and I plan to use adjust_batch() from MMUPHin to correct for technical bias. To handle the sparsity of my matrix (factually, few genera and samples available), I first applied cmultRepl() (GBM) to impute zeros, which returns proportions. Then, I applied adjust_batch() with zero_inflation = FALSE, as follows:

tmp ← cmultRepl(abd_gene_level,z.warning = 1, z.delete = FALSE)

fit_adjust_batch ← adjust_batch(feature_abd = t(tmp),
                                batch = “expr_run”,
                                covariates = “condition”,
                                data = metad,
                                control = list(zero_inflation = F,
                                diagnostic_plot = “mmuphin_diag.pdf”,            verbose = TRUE))

Then I plan to apply a CLR transformation for downstream analyses.

My questions are:

  1. Is it a valid approach to impute zeros first with cmultRepl and then apply adjust_batch() with zero_inflation = FALSE, or is it necessary to let the function handle zero-inflation internally to obtain proper batch correction estimates?

  2. Can this same workflow (cmultRepl → adjust_batch() → CLR) be applied to functional abundance matrices, such as KO predictions from PICRUSt2, to correct for batch effects between sequencing runs?

Thank you in advance for your guidance.