picrust2 normalized table

[Jun-Zhu-stu]

I found the abundance ASV performed difference significant when you normalized by marker gene copies(After this step, ASVs show a difference in each sample.). Likewise, it would perform a big difference on the predication of pathways. So the difference of pathways seem to be caused by the abundance of ASVs after normalization. I want to know whether I need to normalize after the normalization of ASVs by marker gene.

[gavinmdouglas]

Sorry I don't think I follow your question - would you mind rephrasing it?

Thanks,

Gavin

[Jun-Zhu-stu]

When we input an OTU table, picrust2 would creat a table in which be normalied according to the copy number of 16S rRNA. The new table represents the abundance of bacteria, which is applied into the prediction of pathways. However, each sample in the new table shows a different bacterial abundance, which could cause the different abundance of pathways. So I think the new table which represents the bacterial abundance, should be normalized to the same bacterial abundance.

[gavinmdouglas]

You can convert either table of abundances to relative abundances depending on your analysis. Certainly normalizing by 16S copy number will change the resulting pathway prediction abundances, but the idea is that 16S copy number corrected data should be less noisy. In practice it's very hard to show which is more accurate, but that's the motivation anyway.

[Jun-Zhu-stu]

Thanks very much.