Hello,
I have been working on a 16S-seq microbiome abundance experiment analysis for quite some time and I still can't decide which database I should use to assign taxonomy. As a background, I recently switched over from QIIME to DADA2 and I am deciding between SILVA, RDP, and the Genome Taxonomy Database (GTDB). I've never heard of GTDB before DADA2 but this database did a great job at annotating most of my ASVs all the way down to bacterial genera.
My question is has anyone worked with GTDB before and can they speak to its accuracy? Mine did a good job of recapitulating the results from QIIME with Greengenes as the database but I'm still not 100% sold because it seems like SILVA and RDP are the most popular at the moment.
Any suggestions welcome, thanks.
I might be missing the essential information on my mobile phone screen, but I to me the Genome Taxonomy Database seems much genome centric. It looks like a great resource, but what makes you believe it would be a well suited source for a 16S phylogeny?
Silva is in my opinion the gold standard, and version 138 is probably published anytime soon.
For my particular dataset (mouse gut microbiome), GTDB did a good job of recapitulating the "keystone" taxa from the original Greengenes-based taxonomy, resulting in similar compositions of Lactobacillus and clostridia/bacilli abundance ratios.
I'm favoring GTDB at the moment because it's doing the best job at giving me annotation up to genus in most samples and there's nothing that suggests the taxonomy is off/wrong. With Silva, there was a greater proportion of ASVs (and a greater proportion of the total abundance per sample) that were unannotated at the family and genus levels.
Thanks for your opinion, Im considering re-running my whole analysis with silva taxonomy out of curiosity to check for discrepancies. My other option would be to go back, get the original ASV sequences that QIIME constructed and assign taxonomy with GTDB and Silva to see whether there are any irregularities.