Rarefied vs Unrarefied OTU Table for Downstream Analyses

[DoH]

Dear QIIME Experts,

Two questions:

1) When performing ALPHA diversity, is it necessary to have all the samples rarefied to the same number of reads (like in BETA diversity)? Can one perform alpha diversity analysis using unrarefied OTU table?

2) For heatmaps/hierarchichal clustering of samples, does it matter if one uses unrarefied OTU table or rarefied OTU? What is recommended?

I'd be pleased to receive your response.

[Colin Brislawn]

Hello,

This is a great question because it's so controversial! Reasonable people disagree on the best way to do this...

Take a look at this thread on the Qiime 1 forums in which I recap this issue, and then the qiime developers and other folks continue the discussion.

https://groups.google.com/d/msg/qiime-forum/PCYGjcLzOm0/Zk6nmjHHDwAJ 

Sophie Weiss a prominent statistician in this field, really says is best: "An important thing to do is make sure your results are robust to normalization technique.  So, verify your results using two normalization methods."

Colin