In order to calculate the Unifrac Distance, you need a phylogenetic tree of the OTUs.

It does not appear that UniFrac is a stand alone program/code any more. You will have to use QIIME, R, mothur, or something like Qiita (link attached). 

Do you have the raw sequence data? If so, it would be easiest to input it into QIIME to calculate and analyze the Unifrac.

https://qiita.ucsd.edu/

Dear Davis，

      Thank you for your kindly help!

       If I got fasta data for each OTU in  OTU_table above, can I do that ? Because we have no powerful computer to calculate raw data. The file of otu_table.biom generated by qiime for next step calculation seem difficult to get.

       And another question, I have put small data volume raw data (18S rRAN gene sequences) into QIIME, but fail to taxonomy. The database format was not collect. What's the  reference OTUs fomat look like? I don't kown how to retrain the RDP classifier and assign taxonomy. The tuorial in QIIME website is not clear.

        Thank you so much!

Best regrads!

Zhenfan Chen

2016-06-25

When I worked OTUs used a French program in MSDOS directly making the calculation that you mention. Years ago, I don't work at dichotomous choices because I discovered that in nature nothing follows patterns of this type. I currently prefer vector analysis, even when working in a hyper space. Anyway, now I work more with the STATISTICA program that is very complete. There perhaps you can find what you want

Zhenfran Chen,

I think I need a bit more information from you.

What kind of data do you have? Cloning/Sanger? Next-generation sequencing? Amplicon? Whole genome shotgun?

How did you generate/cluster your OTUs?

Assuming you picked OTUs in QIIME, you can create a fasta file of your OTUs using the pick_rep_set.py script (http://qiime.org/scripts/pick_rep_set.html). Then you can align your OTUs with align_seqs.py (http://qiime.org/scripts/align_seqs.html). Now, the align_seqs.py will require a reference alignment; the default is the Green Genes 16s alignment. You will then need to filter the alignment. When the alignment is complete, use make_phylogeny.py (http://qiime.org/scripts/make_phylogeny.html) to infer the phylogenetic tree you will need to calculate Unifrac.

I wouldn't assign taxonomy to the raw data. I would use the OTU rep set: it will be smaller and take less time. The default database should be formatted correctly. There should be no need to retrain the RDP-NBC unless you are using a novel database. I have it is easiest to use the uclust function in QIIME to assign taxonomy and to run the RDP-NBC independently from the command line. 

Dear Davis,

     My data were 18S V4 amplicons of fasta format sequenced by illumina. With your help, I have got the files of seq_outs.txt and rep_sets.txt. But , for next step by assign_taxonomy.py , it's difficult to make two flies for id_to_taxonomy_fp and reference_seqs_fp, though I known what's the exact format  of files should be.For NCBI, the files were too large to use. And for Silva database of  SSU, my trouble is how to make the files from Silva database (I don't konw which files I should download)so that can be use by qiime script, assign_taxonomy.py for metris of '-t' and '-r'.

   Thank you very much!

Best regards!

Zhenfan Chen

2016-06-30

Zhentan Chen,

Hmmm, the SILVA files should be in the correct format already. The current database is in ARB format to make it smaller. I'll look into it. 

If you are running into computational problems, you can try using the online version of Silva (https://www.arb-silva.de/ngs/) or the online version of RDP-NBC (https://rdp.cme.msu.edu/classifier/classifier.jsp) to assign taxonomy. SILVAngs will re-cluster your OTUs, however. THE RDP-NBC will not. You could also try to use the free tier of Amazon's computing cloud or apply for a grant from them (https://aws.amazon.com/).

You do not have to assign taxonomy to build an alignment or calculate the Unifrac distance.

~William

Dear  William,

        I'm really appreciate for your kindly help!  Your advices are very useful to me, and I will try and do it.

        Thank you again!

Best regards!

Zhenfan Chen

2016-07-05

Thank you for all. Every day it is possible to know some thing new, or to have new tools in hand.

Antonio López