I have never used mothur pipeline, but maybe if you use some "OTU picking" with a reference database it will decrease the time you need to run this. Sorry for not be more helpfully.

Dear Do Ngoc duy,

If you want to be able to compare the 16S rRNA genes from your 160 samples, yes you have to analyze them altogether and build a single distance matrix.

Clustering a large number of sequences in mothur with the average neighbor algorithm can take some time. In the latest version of mothur (v1.39), a new clustering method called OptiClust was introduced (for more details see: http://msphere.asm.org/content/2/2/e00073-17). It is accurate and much faster than the average neighbor method.

So if you don't use the latest version of mothur, moving to v 1.39.5 (https://github.com/mothur/mothur/releases/tag/v1.39.5) and using opticlust for clustering (https://www.mothur.org/wiki/Cluster) might solve your problem.

Best,

vincent

--

Hi Vincent Hervé,

Thank you, Yes, I run both options (opti + mothur latest version) as your suggestions but it did not work well. The cluster.split command took more than 600 hrs in our server, and could not give me any results. The temporary files (50 files) reach to 50 Gb each  and it is still running.

I am trying to look back the command and increase some filtering threshold to reduce the numbers of unique sequences but I am not sure if it is a good option.