Loss of feature with Maaslin3

Downstream analysis and statistics MaAsLin
1

Hi,

I am running maaslin3 on my 16s data and I have two treatments; CTRL Vs TWD supplemented with either Kombucha or water. Im running these treatment separately. I did maaslin2 last year which revealed relatively more taxes than maaslin3 and now it has actually changed the story I was trying to explain. I understand that Maaslin3 takes into account both abundance and prevalence and any feature being just pravelant or abundant is thrown away. Right?. Im using pretty much default parameters listed below. Is there anything you recommend?

fit_data = maaslin3(
input_data = taxa_filt_i_w,
input_metadata = df_input_meta.i.w,
output = “sample_run”,
fixed_effects = c(“Drink”),
normalization = “TSS”,
max_pngs = 250,
standardize = TRUE,
cores = 1,
max_significance = 0.1
)

2

Hi,

The parameters you’ve set all look reasonable to me. MaAsLin 3 will tell you whether the abundance or prevalence association is significant separately (individual q-values) or together (joint q-values), so it doesn’t throw away features that just have abundance or just have prevalence associations. We’ve found the MaAsLin 2 and 3 outputs to generally agree, so I’d be surprised if very significant results before became insignificant, especially with a model this simple.

WIll