A problem in clustering

[hyungz]

When I clustered sequences into OTUs at a distance cutoff of 0.03 with average-neighbor method and obtained representative sequences from each OTU with the command “get.oturep”,

I frequently encountered the case that the similarities between the representative sequences were above 97%.

I am afraid that this indicates that the number of OTUs was overestimated.

How can I reduce this phenomenon?

Thank you.

hyungz.

[pschloss]

There is no perfect method for clustering sequences. If you look at our 2011 AEM paper you’ll see that of all the methods, average neighbor is the best.

[hyungz]

Thank you for your kind reply. I understand.