Hi QIIME devs,
Would it be possible to incorporate a plug-in in the future so that we can run DESeq2 or metagenomeSeq fitzig for differential abundance without having to export the feature table to QIIME1 or R? It would be great to be able to run those differential abundance algorithms in QIIME2.
Thanks!
3 Likes
Hi @jroth002! We have an open issue to support additional normalization strategies in QIIME 2. I don’t have an ETA for when this functionality will be available (it will either be added to the q2-diversity plugin, or as a standalone plugin). We’ll follow up here when the functionality is available in a release!
If you know of anyone interested in helping us with this feature, we’d love to see a new Community Plugin for normalization strategies! We have a plugin developer guide to help with creating a QIIME 2 plugin. Thanks!
1 Like
Along with that would be the normalization procedure that was in QIIME 1.9 http://qiime.org/scripts/normalize_table.html
That would be great!
I’m going to reply to that open issue also, but the normalized dataset should be used for all diversity analyses including alpha (those that incorporate abundance such as Shannon), beta and differential abundance testing. The open issue makes it sound like you should incorporate the option for beta diversity testing and that alone.
If you are doing analyses on presence-absence data that is fine without normalization, but anything that uses a sequence count as a representative of ‘abundance’ should be normalized. Recent paper I just saw (not peer-reviewed yet): https://www.biorxiv.org/content/early/2017/12/11/231878, and I know that McMurdie & Holmes feel that rarefaction should not be used at all for any diversity analyses: https://github.com/joey711/phyloseq/issues/603#issuecomment-219180008
2 Likes