ASVs annotation using classify-consensus-blast

[Yogesh_Gupta (Yogesh)]

Hi,

I am doing ASVs annotation with classify-consensus-blast using below command and I got taxonomy with confidence ranging from 0.0 to 1. Which confidence value can be considered as correct annotation? Can I consider annotation with confidence above 0.50 for downstream analysis or I do need to mention some parameter to remove the false annotation?

qiime feature-classifier classify-consensus-blast --i-query rep-seqs.qza --i-reference-reads silva_132_99_16S.qza --i-reference-taxonomy raw_taxonomy_silva_132_99_16S.qza --p-perc-identity 0.8 --p-strand both --p-evalue 0.0001 --o-classification taxonomy-blastconsensus-silva.qza

Many thanks,
Yogesh

[Nicholas_Bokulich]

Hi @Yogesh_Gupta,
classify-consensus-blast reports a consensus score, not a confidence score (no matter what the label is, it is the fraction of hits that match the given taxonomy for this method). That is how to interpret the result, not as confidence.

@Yogesh_Gupta

Which confidence value can be considered as correct annotation? Can I consider annotation with confidence above 0.50 for downstream analysis or I do need to mention some parameter to remove the false annotation?

That is up to personal choice. The default settings were chosen because they performed best according to our published benchmarks. But what threshold you want to use for classification is up to you.

Good luck!