Zero filtering of custom Log Ratio data for MaAsLin3

Downstream analysis and statistics MaAsLin
1

Hi,

I have Log Ratio data looking like this:

image
image553×168 5.72 KB

I want treat as zeros all values below -12.20607 (log(0.000005)) .

But when I run the command below, I see that no values have been filtered out, thus I am getting no prevalence results.

image
image442×348 8.12 KB

Here is the result table where you see all values are non zero (1492) and that prevalence modelling is not working:

image
image978×415 15.5 KB

What can I do to solve this?

Thanks,
Victor

2

Hi Victor

Can you go to your output directory and paste the first few rows and columns of the TSVs in features? That would help me get a sense for where things are not going as expected.

Will

3

Hi Will,

filtered_data.tsv

image
image857×189 11.2 KB

data_transformed.tsv

image
image922×230 13.5 KB

data_norm.tsv

image
image832×225 11.9 KB

Thank you for the support,
Victor

4

Hi Victor,

It looks like we had a bug in the code for normalization='NONE'. Can you try installing again from here:

devtools::install_github('https://github.com/biobakery/maaslin3', ref = 'devel')

And letting me know if it works?

Will

5

Hi Will,

Yes, it works now!

Thank you very much,
Victor