BLAST consensus question

[SoilRotifer (Mike Robeson)]

Hi @Newt,

I suspect, that part of the reason why classify-consensus-blast is ultimately choosing Dendrodrilus rubidus, has to do with the --p-min-consensus setting. If you read the help documentation by running:

qiime feature-classifier classify-consensus-blast --help

You'll see that the help text for --p-min-consensus says:

--p-min-consensus NUMBER Range(0.5, 1.0, inclusive_start=False,
    inclusive_end=True)   Minimum fraction of assignments must match top hit
    to be accepted as consensus assignment. [default: 0.51]

This setting, as well as what sequence data is ultimately included within your curated reference database, can alter the taxonomy consensus assignment.

@Newt

I'm a little concerned about using a species level classification if NCBI blast is identifying multiple species with 100% identity. I'm wondering if I should move classification down to a different level if the sequence belongs to multiple species.

This is quite common with short-read amplicon data. Which is why there are a variety of consensus approaches (vsearch & BLAST), and other taxonomy assignment tools (scikit-learn, RDP,...) out there.

Also, I recommend trying this new RESCRIPt approach for generating a reference database from GenBank data.