Hi, I am running MaasLin2 with following lines of code using data.csv (4.8 KB) and metadata.csv (9.2 KB) files. But my machine automatically freezes after few minutes. I have run other fies with same fomat coming from other datasets. In those cases MaasLin2 is running fine.
Can you please tell me why I am facing this issue specifically for this dataset and this set of files?
#run maaslin2 without covariates
input_data <- as.data.frame(read_csv("data.csv"))
fit_data_crude <- Maaslin2(
input_data, input_metadata, 'output_crude',
fixed_effects = c('type'),
reference = c('comorbidity,NGT'),
min_abundance = 0.0,
min_prevalence = 0.0,
normalization = "NONE",
analysis_method = "CPLM")
I don’t know specifically why it’s freezing, but your
data.csv file doesn’t seem to have any species information in it, just two columns labelled ID and value. Try examining the input data in the tutorial and get your data to look like that.
Thanks @andrewGhazi for your response. In this case I am using MaasLin2 not for species/function purpose rather i have a different aim to use it. However, please note that same format I successfully ran for other multiple datasets where the sample names started with ERR in place of SAMEA.
I tried running your command and it completed for me, though it looks like the CPLM function isn’t able to successfully optimize the likelihood function. I tried
input_data |> inner_join(input_metadata) |> cplm::cpglm(value ~ type, data = _)
(which should be pretty close to what Maaslin2 is running under the hood) and got a false convergence warning.
Given that you only have a single outcome feature to look at (which might not even be a microbial species abundance?), you might want to eschew using Maaslin2 at all and use a simpler analysis method. Looking at a beeswarm of your data, it doesn’t look like there’s much of a difference between the two groups: