# NCBI or SILVA for curating a custom 16S rDNA reference database? **URL:** https://forum.qiime2.org/t/ncbi-or-silva-for-curating-a-custom-16s-rdna-reference-database/34286 **Category:** General Discussion **Created:** [August 12, 2026, 6:11am UTC](https://forum.qiime2.org/t/ncbi-or-silva-for-curating-a-custom-16s-rdna-reference-database/34286 "2026-08-12T06:11:29Z") **Posts on this page:** 1 **Showing post:** 4
### Author: ![SoilRotifer](https://forum.qiime2.org/user_avatar/forum.qiime2.org/soilrotifer/32/21071_2.png) [@SoilRotifer](https://forum.qiime2.org/u/SoilRotifer) #### Post date: [September 12, 2026, 3:23pm UTC](https://forum.qiime2.org/t/ncbi-or-silva-for-curating-a-custom-16s-rdna-reference-database/34286/4 "2026-09-12T15:23:35Z")
> [@deClare125](#): > > I've been encountering some errors that may prevent me from continuing to create a custom reference database, but I'll elaborate on that later/ elsewhere if I don't manage to troubleshoot on my own. Please let us know if you are unable to work through it. > [@deClare125](#): > > Should I be paying attention to the difference between 16S rRNA and 16S rDNA within these reference datasets? T Nope. Actually, there is no such thing as rDNA. Although I must confess, I use that term too. I use it as a shorthand to mean '16S rRNA gene'. Though that makes some folks unhappy. 😬 SILVA pulls DNA sequences from several online databases and resources and then transcribes them to RNA sequence. The reason for this, is that they curate the rRNA alignment by using secondary structure information. The only proper way to do this is to use the transcribed RNA sequence. You can confirm this yourself, if you go to the online SILVA website and look up the various sequences (or look through the downloaded files) you'll see that the GenBank, Accession or other identifier is used as the sequence ID. If you search for that on GenBank, you'll see that the reference sequence is likely in DNA form. FYI, SILVA is more than just a sequence repository, it collates a lot of information from a variety of resources, such as GenBank, DSMZ, etc. SILVA is truly great! You can read more about it [here](https://doi.org/10.1093/nar/gkaf1247). Also, SILVA just released pretrained classifiers for version 144 [here](https://forum.qiime2.org/t/silva-138-and-144-classifiers-are-now-available-for-qiime-2-2026-7/34321). We are in the process of updating RESCRIPt to fetch the [144 database](https://github.com/bokulich-lab/RESCRIPt/pull/264). > [@deClare125](#): > > If I download data from MIDORI or Eukaryome that specify 16S rRNA, will I need to reverse transcribe all of my data to rRNA, or all the reference sequences to rDNA in order to assign taxonomy? The reference data is in DNA form. > [@deClare125](#): > > When I downloaded NCBI data, I was careful to use --p-query "txid33208[ORGN] AND (16S OR 16S rDNA OR 16S ribosomal DNA)", I often like to include 18S as many of the 16S rRNA gene primers will also hit the 18S rRNA gene too. These two genes are also known as small sub-unit (SSU genes), as they are homologues and explains why they are within the same database / alignment. The 23S rRNA gene and 28S rRNA gene are large sub-unit (LSU genes) and they are curated within a separate alignment. In a nutshell, I like to have outgroup taxa to more clearly identify, and remove, non-microbial sequences. In many cases, if a reference database does not contain any outgroups you can mis-identify non-microbial organisms as microbial. Also, have you tried pulling the reference data from GenBank as outlined [here](https://forum.qiime2.org/t/using-rescript-to-compile-sequence-databases-and-taxonomy-classifiers-from-ncbi-genbank/15947)? Also, keep in mind that different reference databases might use different taxonomic schemas, so you might run into inconsistencies and might need to correct these. -Cheers! --- _[View the full topic](https://forum.qiime2.org/t/ncbi-or-silva-for-curating-a-custom-16s-rdna-reference-database/34286)._