Classification thresholds for ranks above phylum level about crest4

[lanzen]

Hi,

To first answer your question, there is no similarity threshol for ranks above phylum. Also, if you want to see what phylum that it assigns your OTUs to without the similarity cutoff, you simply can unactivate it usign the option -f (or --nofilter). Because I assume that this is a result from CREST v3 and not CREST4 although you have filed the issue here? UNITE was never made available for CREST4 and unfortunately I have stopped maintaining it, but I could try to find time early next year to provide a version for CREST4 based on the latest release.

The sequences that are assigned to only supergroup and is not called "unknown" something (when not using -u, because then all "unknown" ranks are removed), they result from either (i) a very low percentage similarity and very dispersed matches, (ii) to one or more matches with severely conflicting taxonomy, or (iii) high similarity to a sequence with very little taxonomic resolution (although all reference sequences that did not have at least phylum rank info were removed if I remember it correctly). I have seen this during testing for PR2 since it is not based on a global alignment tree, as opposed to SILVA. Can you send me the OTU sequence so that I can look at it? If it is the later, I should revise the database to remove the problematic reference sequence.

Best,
Anders

[xapple]

If this is resolved, I will close this issue. If there are more problems, don't hesitate to reopen.

It seems to me this concerns crest3 and not crest4.