TSS normalization vs qiime2 relative frequency

Hello everyone,

I have one question about the difference between TSS normalization and qiime2 relative frequency.

I feel a little confuse because in literature on papers for example here doi:10.1080/19490976.2023.2244139 and others papers, explain that after the TSS normalization the feature table become relative abundance table. Although when we use the command qiime2 relative frequency as result is a relative abundance table.

So here is my question There is any different between TSS and qiime2 relative frequency?
Or is the same approach?

I am not familiar with statistics consent and formulas.

many thanks


Yes, these are the same!

TSS is Total Sum Scaleing. This means that raw counts are converted/scaled based on the total sum of counts in that sample.

This is a fraction, i.e. relative to 1. This is also called a proportion.
If you multiply it by 100, it becomes a percent, i.e. relative to 100.

This paper has a good overview of other methods, like CSS and CLR. While I can't comment on their new method, I like how they describe the existing methods and their issues.



Thank you for you response and for the paper!

Many thanks