Should I need to do normalization to remove batch effect when I have different studies with LEfSe analysis?

Question

[Qiyan]

I have different bacteria studies, and the input of LEfSe just demands relative abundace. You know, relative abundance is a kind of normalization, but I don't sure it is enough to remove batch effect. Should I do normalization such as combat or limma, or I just merge relative abundance of different studies as input?

[Qiyan]

I found a paper about how to do normalization about microbiome studies, but I not sure the method can be used for LEfSe. Correcting for batch effects in case-control microbiome studies, link: https://doi.org/10.1371/journal.pcbi.1006102.