Cluster.split or phylotype binning


What are the practical differences in the outputs from these two ways of generating a shared file. I have tried cluster.split, but run into trouble because of very large files, but the phylotype worked fine. Can I use these data in a further analysis as usual, or is there anything I havn’t understood about major differences?


Are you using opticlust? I haven’t needed to use cluster.split since that method was published (and I on occasion have huge very diverse datasets). I’ve never been a fan of phylotype binning because that tells you about the humans who work on bacteria rather than the bacteria themselves (what gets named is completely dependent on a human deciding its worth working on)


Thanks for your reply! No, I’ve not tried opticlust. I’ve just been following the mother MiSeq SOP with cluster.split. Have I understood it correctly that I first have to run dist.seqs and the cluster with ‘opti’ as method?


@toneta can you make sure you are using the latest version of mothur (v.1.41)? cluster.split is a de novo clustering approach that should generate a shared file that resembles what you’d get with running cluster. The phylotype command generates OTUs based on their assignment to groups in taxonomic levels that you specify in the command. There will be many more de novo OTUs than phylotypes because the de novo OTUs are more specific than anything you’ll get out of phylotype.

The default clustering algorithm in cluster and cluster.split is opticlust.


This topic was automatically closed 10 days after the last reply. New replies are no longer allowed.