qiime diversity core-metrics-phylogenetic: singletons in rarefied table

[dancurtis87 (Daniel Curtis)]

Good morning ,

I am seeking clarification on the appearance of singletons in my feature table after using the core-metrics-phylogenetic pipeline.

I am working with a feature table prepared from two sequencing runs(Run A & Run B). Each run was performed with same sequencing protocol but at different facilities. After denoising each run with DADA2 there are two singletons present in feature table from Run A and none in the feature table from Run B. Based on posts on the forum, I understand these can be introduced during the merging of forward and reverse reads.

After merging the individual denoised data with qiime feature-table merge, the resulting feature table contains ~ 11,809 ASVs (n = 68 samples at this point). However, after running this table through qiime diversity core-metrics-phylogenetic (–p-sampling-depth 10000), the rarefied table (n = 66 samples) contains 11,138 ASVs, of which ~ 900 are singletons.

I don’t quite understand how these singletons appear in the feature table and would appreciate any insight.

Thank you,
A quarantined qiime-er.

[jwdebelius (Justine Debelius)]

Hi @dancurtis87,

When you do rarefaction, you are resampling your table without replacement until you get to a specified depth.

If I have a feature that has a frequency of 1/10000 (1e-5) in three samples, and then I resample to 10000 sequences, that feature may only show up once instead of three times, making it a singleton.

Typically, rarefaction analysis is used for diversity (unless you’re using decoide and/or breakaway, although Im not sure if q2-breakaway is still up to date.). I think Weiss et al motivate it well.

Best,
Justine

[dancurtis87 (Daniel Curtis)]

Thank you for the clarification @jwdebelius!