Hi,

If you have used the same OTUs for the different methods, then the answer is "yes". In fact, it is a common practice to infer phylogenetic trees using different programs and/or approximations, e.g. RAxML vs PhyML, Bayesian trees vs. Maximum Likelihood vs. Neighbour Joining, etc.

If you haven't used the same set OTUs then you can still compare phylogenetic trees for the set of shared OTUs. On this case, you should prune trees to keep only common OTUs prior to any comparison. Although you can perform such comparison, I'd not advise to make comparison (unleast it is strictly neccesary) for different set of OTUs since other OTUs may impact on the reconstruction of trees.

I hope this helps.

Hi,

If you have used the same OTUs for the different methods, then the answer is "yes". In fact, it is a common practice to infer phylogenetic trees using different programs and/or approximations, e.g. RAxML vs PhyML, Bayesian trees vs. Maximum Likelihood vs. Neighbour Joining, etc.

If you haven't used the same set OTUs then you can still compare phylogenetic trees for the set of shared OTUs. On this case, you should prune trees to keep only common OTUs prior to any comparison. Although you can perform such comparison, I'd not advise to make comparison (unleast it is strictly neccesary) for different set of OTUs since other OTUs may impact on the reconstruction of trees.

I hope this helps.

I do support both the answers provided by Dr. Salvador Capella-Gutierrez and Dr. Duo Peng. The choice depends on the objective of your analysis. Depending on the need and core objective of the study you should opt your strategy. But finally whatever tree you get it is just the "best guess" and to justify that you seek statistical support.

Dear Salvador, Duo and Amit thanks for your answers

My purpose is to identify monophyletic clades "matching" in each method can make a selection of OTUs that are in all monophyletic clades "matching" of all methods and discard any that are not.

The idea is classify a gene ortholog.

Regarding my question

I can not find information that tells me if it is possible to compare trees that have been inferred with different methods. And if it is possible, why not?

The closest thing I've found are the supertree

Could you recommend any book or paper?

"In each method used the same OTUs "

ok, but for which taxonomic group?

Maybe these could be good options for your purpose:

http://bioinformatics.oxfordjournals.org/content/22/1/117.full.pdf

http://bioinf.nuim.ie/wp-content/uploads/2011/10/TOPD-FMTS-A-new-software-to-compare-phylogenetic-trees_2007_Bioinformatics.pdf

Thanks Gustavo

Regarding to the methods to compare phylogenetic trees, a couple of years ago, a labmate published TreeKO [Marcet-Houben M, Gabaldón T. TreeKO: a duplication-aware algorithm for the comparison of phylogenetic trees. Nucleic Acids Res. 2011 May;39(10):e66. http://www.ncbi.nlm.nih.gov/pubmed/21335609] which is aware of duplication & speciation event in the trees being compared but it also consider for any comparison the common set of sequences among the trees being compared.

As Duo Peng mentioned, there is just one true phylogeny for your set of sequences/species and with current methods we have trees which approximate to this true phylogeny but we have no complete certanty about that.

I'd like to recommend a couple of paper regarding to the different approximations to reconstuct phylogenies and regarding to phylogenomics.

Delsuc F, Brinkmann H, Philippe H. Phylogenomics and the reconstruction of the tree of life. Nat Rev Genet. 2005 May;6(5):361-75. http://www.ncbi.nlm.nih.gov/pubmed/15861208

Snel B, Huynen MA, Dutilh BE. Genome trees and the nature of genome evolution. Annu Rev Microbiol. 2005;59:191-209. http://www.ncbi.nlm.nih.gov/pubmed/16153168

I hope this help,

In case you are interested, the web-based tool for the paper: A novel algorithm and web-based tool for comparing two alternative phylogenetic trees is now at:

http://www.mas.ncl.ac.uk/~ntmwn/phylo_comparison/pairwise.html

Thanks Salvador and Gustavo

I'll read the papers. Compare2Trees online tool I feel great, I will read the paper also.

Thanks again

The simplest indirect method is PATRISTIC (Fourment et al), and this has the advantage of helping you further 'explore' the data.

”A much more specific question would help a great deal here. What type of data are you using? What are the different methods? And what is the hypothesis - what question are you trying to answer?”

let's bet that he hasn't got any...

Dear Patrice Corneli, thanks for your reply

The data I use are nucleotides.

The methods that I want to compare are NJ, maximum parsimony, maximum likelihood and Bayesian.

The purpose is methodological, in my view, the best trees of each method are more approximate to the real, and if you only consider as OTUs closely related to which are grouped into a monophyletic clade in all methods, the report of my results can be more accurate.

It is important to clarify that my study does not attempt to infer the evolutionary history itself, the purpose is to predict the function of genes and classify them into different classes.

Best Regards