Questions about PCoA analysis to analyse the clone library data

[8409...@qq.com]

Hi,

   I would like to ask some questions about PCoA Analysis(weighted unifrac) for clone library data.

   I have about 1000 sequences of the functional gene in 10 samples, every sample is about 100 sequences. I don‘t know my data whether or not fit for this analysis and which detail I need to notice, for instance, whether or not the number of my data is too few to analyze by QIIME (only 1000 sequences all 10 samples), because generally QIIME is used to analyzed huge data, such as metagenome, and other detail. I need some suggestions.

   I analyzed the gene sequence as follow, I removed the primers, translated DNA sequences to the amino acid, then aligned by clustalX and DOTUR for OTU. I created an OTU table in Excel and transformed the format to biom file by mothur program. I made a tree file by using Phylip and I made a mapping file.

   One question is about tree file. I really want to know what kind of tree should I build. Could I add an outgroup in the tree file just like Fast Unifrac? If I added the outgroup, do I need to add outgroup OTU to the OTU table? I have tried it if I add outgroup to phylogenetic tree following the same step like Fast Unifrac, but not add the name of outgroup to OTU table. Unfortunately, I can't get the result if I did like that.

   Anyway, now, if I had the files, OTU table, tree file and mapping file, so I continue doing the PCoA analysis in Qiime.

   First, following the Qiime scripts, I did the beta_diversity.py, used the Single File Beta Diversity (phylogenetic) Method. After that used principal_coordinates.py script and PCoA (Single File) Method to do PCoA Analysis. Finally, I make 2D PCoA plot images use Default Example way and make_2d_plots.py, got the result figures.

   But in the step, I get some warning, as below:

(qiime1) duanweixiangdeMacBook-Air:~ duanweixiang$ beta_diversity.py -i /Users/duanweixiang/Downloads/2017326.1.biom -m weighted_unifrac -t /Users/duanweixiang/Downloads/example_fasta_after_split_libraries.tre -o /Users/duanweixiang/Downloads/

(qiime1) duanweixiangdeMacBook-Air:~ duanweixiang$ principal_coordinates.py -i /Users/duanweixiang/Downloads/weighted_unifrac_2017326.1.txt -o beta_div_coords2017326.txt

/Users/duanweixiang/miniconda3/envs/qiime1/lib/python2.7/site-packages/skbio/stats/ordination/_principal_coordinate_analysis.py:107: RuntimeWarning: The result contains negative eigenvalues. Please compare their magnitude with the magnitude of some of the largest positive eigenvalues. If the negative ones are smaller, it's probably safe to ignore them, but if they are large in magnitude, the results won't be useful. See the Notes section for more details. The smallest eigenvalue is -0.0610505242062 and the largest is 1.29980252151.

  RuntimeWarning

(qiime1) duanweixiangdeMacBook-Air:~ duanweixiang$ make_2d_plots.py -i /Users/duanweixiang/Downloads/beta_div_coords2017326.txt -m /Users/duanweixiang/Downloads/Category.txt -o /Users/duanweixiang/Downloads/

/Users/duanweixiang/miniconda3/envs/qiime1/lib/python2.7/site-packages/matplotlib/collections.py:590: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison

  if self._edgecolors == str('face'):

(qiime1) duanweixiangdeMacBook-Air:~ duanweixiang$ 

   I can't understand the warning in red color. Why did they appeared, what they mean? how to analyze the warning and avoid them?

[Greg Caporaso]

Hello,

Replies are inline below:

too few to analyze by QIIME

QIIME will work ok on data sets of this size.  

I really want to know what kind of tree should I build?Could I add a outgroup in the tree file just like Fast Unifrac?

In QIIME 1, you can use either rooted or unrooted trees with UniFrac. So, the unrooted tree that you've generated would be ok, but you can also perform midpoint rooting or outgroup rooting. If you perform outgroup rooting, you do not need to include the outgroup in the OTU table. If you tried that and received an error message, can you please post the command that you ran and the error that you received?  

I can't understand the warning in red color. 

These warnings can be safely ignored, they won't impact your results. The first warning message includes a description of when it is safe to ignore. The second is a warning that is more important for developers than for users (it's telling us that an API we're using is going to chance). 

[gaojun1...@gmail.com]

Hello Greg Caporaso,

 Thank you for your advice!

I will follow your advice,then feedback to you.

best,

Duan