UniFrac and denovo alignment

[Hazem]

Hello All,

I am interested in using Unifrac metric (weighted and unweighted) for my beta diversity analysis. The usual catch is that I am using a non-marker gene and there is no good reference alignment for it, since it is usually sequenced using several primers amplifying several regions in different studies.

My question is it sound to do my own denovo alignment from the rep set? I can trim and filter the alignment after that, and then generate a maximum likelihood tree that can be used for Unifrac analysis. I have about 17,000 OTUs in my rep set.


PS. The unweighted unifrac PCoA plot doesn't differ much from the bray curtis one. More analysis is needed to see the biological effect though.


--
Cheers,
Hazem

[zech xu]

Hi Hazem,

You are perfectly right that you have to build alignment and tree on your rep set. You can use the QIIME commands align_seqs.py, filter_alignment.py, make_phylogeny.py to do that.

[Hazem]

Hello Zech,

I didn't use the align_seqs since I aligned with MAFFT. Muscle would not work on a denovo alignment of ~17,000.

Referring to this post, I just have a question regarding the use of the filter_alignment script (I am using QIIME 1.8). There appears to be a bug in the script when I use the -r option. So I am just using -g and -e to filter gaps and highly entropic bases at 5%. Do I have to put the arguments in a certain order or use the entropy threshold of 0.0005 as it is mentioned int the post? My alignment lengths were almost 800bp, and they got filteredis filtered down to 300bp, which is close to the amplified gene length. I just put -g 0.9 -e 0.05 as gaps are removed first. I chose the entropy of 0.05 based on the sequence length.



--
Cheers,
Hazem

[Antonio González Peña]

filter_alignment.py is meant to be use when you align against a reference (alignment/mask) and the reference has a lot of gaps => mainly meant for pynast. Thus, I'm not sure if you need it for your use case.