Kraken2/bracken with 16S sequencing downstream analysis

Question

[Isidro]

Hello everyone,

I’m working on a 16S rRNA microbiome project using Kraken2 with the SILVA database for taxonomic classification. After classification, I apply Bracken to estimate species-level relative abundances, and I’m using the alpha_diversity.py script from KrakenTools to compute alpha diversity indices (Fisher, Shannon, Simpson, Inverse Simpson, and Berger-Parker).

I chose Kraken2 over tools like DADA2 or QIIME2 based on a recent benchmark study suggesting that Kraken2 offers higher accuracy for 16S data in certain contexts.

My question is about how to handle downstream analysis, especially:

Since Kraken2 doesn’t generate OTUs or ASVs, I’m unsure whether rarefaction is still needed before calculating alpha diversity.

My samples range between 25,000 and 50,000 total reads. Can I trust the diversity metrics computed by KrakenTools without rarefaction or normalization?

My preprocessing pipeline so far only includes Fastp for quality filtering. Should I include additional steps (e.g., chimera removal, filtering low-abundance taxa)?

Would it be more appropriate to switch to a DADA2/QIIME2-based pipeline for downstream diversity analysis?

I’d really appreciate insights from anyone who has worked with Kraken2/Bracken for 16S data, especially regarding best practices for diversity analysis and normalization.

Thanks in advance!

Answer 1

[andres.firrincieli]

That paper (https://www.nature.com/articles/s41598-023-40799-x) conflates a taxonomic classifier (Kraken2) with a broader analysis platform (QIIME2), which is primarily used to generate ASVs and ASV abundance tables.

A more appropriate comparison would be between Kraken2 vs QIIME2's naive Bayes classifier.

You can still use a DADA2-QIIME2 workflow to generate ASVs and abundance tables, and then apply Kraken2 separately for taxonomic classification.