Help on normalization

Normalization is usually required prior to any analysis. There are many type of normalization methods, e.g., rarefying, CSS, regularized log transformation,etc. Is there any recommendation?

A recent paper [Sophie Weiss,2017] did simulation and suggested rarefying. But if I use metagenomeSeq or DESeq/DESeq2, should I use rarefying or their own suggested one?

In Qiime2, rarefying is applied to alpha/beta diversity calculation. Is there an function to apply rarefying on OTU?

I am a beginner of microbiome data analysis and really appreciate if there is a guideline for the whole preprocessing and differential analysis…



Hi @lindd, thanks for reaching out!

Great question! I would suggest contact the support forums for those tools and ask your questions there — this is the support forum for QIIME 2, so we aren’t really in a position to provide that kind of support for third-party tools.

It sounds like you are referring to diversity core-metrics / diversity core-metrics-phylogenetic — this is a convenience method that actually rarefies your feature table (OTU table), and then calculates alpha and beta diversity metrics using the rarefied table. You can also rarefy a table independently of these methods, and calculate alpha and beta metrics on their own.

I would suggest working through our “Getting Started” doc - then work your way through the various QIIME 2 tutorials to become comfortable.

Let us know if you get stuck or have any questions!

Happy QIIMEing! :sun_with_face:

1 Like

Thanks for informative reply.


1 Like

This topic was automatically closed 31 days after the last reply. New replies are no longer allowed.