Hi,

I am seeking opinions on the best way to determine OTU identity using Blastn. Would the best way to identify an OTU be to blast the OTU to the full nr/nt Nucleotide collection or to blast your OTU to a reduced database made up of sequences that target your organism of interest (for example, the organism to which you designed the sequencing primers for)?

An example to clarify the question:

Scenario 1: You blast your OTU to the full nr/nt Nucloetide collection and get the following best hit: Homo sapiens with an E-value of 3e-175.

Scenario 2: You then blast the same OTU to a database made up of only reads for your organism of interest (Pan troglodytes). The hit you get is for Pan troglodytes with an E-value of 1e-40.

Both E-values suggest that the hit is significant and not due to random chance (http://sfg.stanford.edu/BLAST.html), but the Homo sapiens result is a better match due to its smaller E-value. However, if you know that you should only have matches to Pan troglodytes, is it justifiable to use the less significant match? Which approach is better?

Thanks in advance for any opinions.