Here's my poster for #ABACBS2023!

We compared different #eDNA taxonomic classifiers across different simulated datasets of 12S and 16S metabarcoding data.

MMSeqs2 generally 'wins' closely followed by BLAST.
With exclusion databases it becomes apparent that standard Kraken2 produces many, many false positives. With some families like snappers, all classifiers can't assign more than ~30% of sequences.

(not shown is COI, where MMSeqs2 and Mothur are working best)