I have two phylogenies, one obtained with ITS and another with trnL. The ITS has 95 taxa and the trnl has 104. Most of them are from the same specimens, but there others that are unique to ITS or trnL. How do I "prune" the taxa that are not present in both phylogenies? I want to compare the two phylogenies through the program Compare2Trees (http://www.mas.ncl.ac.uk/~ntmwn/compare2trees/index.html)
(BTW, I learned about this program here in this forum!)
Many thanks and looking forward to hear from you.
The simplest solution would be to remove the taxa which are not in the other matrix before the analysis, either manually, or with PAUP for instance (option delete taxa).
If you want to use the trees that you have already produced, you can drop taxa from a tree using the command drop.tip as implemented in the package ape in R. I'm sure there are other programs that can also drop tip if you're not familiar with R (but I don't know them).
Two programs that can prune taxa are Phyutility and Mesquite. However, there is always a risk that the inclusion and exclusion of taxa may affect the estimation of the phylogeny. So it is probably better to do the analyses with the same set of specimens
You could also just put them in the same order with
Sample Name NNNNNNNNNNNN
for the sequences you don't have. Interleave it,in the nexus file, but make each gene it's own partition. Then run a concatenated analysis. RAxML should be fine, but Bayesian stuff can usually tolerate missing data fairly well as long as it's not too much/systematic in my experience .
If your really bold you can try for a species tree although the missing data may be an issue. I have blocks I could post somewhere if you don't know how to set it up in Mr. Bayes
Maybe I am missing the point here, but if the idea is to infer a species phylogeny and not just compare the gene phylogenies, then the state-of-the-art analysis is a multispecies coalescent-based one (using e.g. *BEAST), under the assumption that the phylogeny is tree-like. The analysis tolerates missing sequences and multiple samples/sequences per species.
From the extensive time I spent looking into the same question, it appears that all methods require the same number of terminal taxa and it's best if they have the same name, all of which has been mentioned above. There are various programs to compare the actual trees.
I also suggest the ape package for R for tree editing (deleting unnecessary tips etc.), but if you prefer a software with GUI you can use Archaeopteryx from the forester project.
https://sites.google.com/site/cmzmasek/home/software/archaeopteryx
It is a java software (meaning cross-platform) and knows a lot about tree editing (and some analysis too!). I really like its features.
I suggest a reconciliation of the gene trees with a common species tree. Then do a comparison of these reconciliation. For smaller trees this can be displayed quite nicely.
Of course, if you do not know the species tree, then finding the species tree will open a whole new can of worms.
Hi, I can suggest to try Mesquite Software v2.6 (Maddison and Maddison 2007). We used it to build the congruency between ITS and EF1 alpha datasets. Details of this study you can find in "Barseghyan GS, Kosakyan A, Didukh M, Wasser SP, Nevo E, Isikhuemhen OS. 2011. Phylogenetic analysis within genera Morchella (Ascomycota, Pezizales) and Macrolepiota (Basidiomycota, Agaricales) inferred from rDNA ITS and EF-1α sequences. In Systematics and Evolution of Fungi. Edit. J.K. Misra & S.K. Deshmukh, Science Publishers, NH, USA. 426 pp." I know that this may not be very modern approach, but it worked quiet well in our case. Good luck. Anush
In my opinion is not a matter of prune or delete some taxa in the tree since the inclusion and exclusion of taxa do affect the estimation of the phylogeny.
I agree with Jesus and Thomas. A basic premise of systematics/phylogenetics is that relationships among taxa are represented by an irregularly-bifurcatng hierarchy. In the absence of some a priori true tree, your best hypothesis of taxonomic relationships is inferred by combining the data you have and analyzing them as a single matrix. Missing data (or taxa) should not be too much of a problem, as long as the missing data are not complementary (taxon a has gene 1 but not gene 2, taxon b has gene 2 but not gene 1, in which case the software may interpret them as being identical due to ambiguity of missing data). What I would do with your data set is take the taxon sample for trnl and include all the sequences of ITS for the same taxa and analyze that under your preferred optimality criterion to get a hypothesis of taxonomic relationships. Then you can use that as a baseline for comparing incongruence or whatever of the individual gene trees.
I would agree that the most up-to-date mathod would be a coalescent analysis like *Beast. If you are interested in making a species tree, you can also try Phylonet - it computes a species tree and all you need are gene trees in the newick format. It can handle missing taxa.
From a little bit older perspective there was also a super-tree (super-matrix) analysis which was often used when there was an issue with uneven sampling (there programs as well for that).
But it depends also a bit on the purpose. You are from botany so perhaps you would like to compare nuclear vs. chloroplast phylogenies to get an idea about hybridization? It is a bit out of date but sometimes still legitimate (there are still publications using that methodology). It may form a first step in the analysis as well. However such studies are mostly done by visual comparison. Once I was thinking to use programs which were developed for comparing host-parasite phylogenies but I never tried it this way.
But again I could suggest you a phylonet program - it can also make a species network (infer reticulations without specyfing them a priori), which would give you information about hybridization. But maybe you should be a bit cautious using only two loci, however if you have a good support in your gene trees you can try.
Good luck!
Dear all,
Thanks to everyone who have answered. I want to obtain a species tree, and before I concatenated both genes, I wanted to know whether there was congruence betweeen ITS and trnL. From this forum and from the literature, I understand that several of the congruence tests are very conservative. Ideally, and when I have more genes, I would like to try *Beast. For the time being, I am going to try with Phylonet.
Best and thanks again to all for sharing your ideas/suggestions.
Alina
- Badges
- Science topic
- Similar topics
- Biological Science
- Zoology
- Ornithology