Since the maximum number of sequences per cluster is higher than the maximum number of species per cluster, i deduce your orthologous clusters contain (in)paralogs. I also assume the clusters are from different families and what you want to infer is the species tree. Let me know otherwise.

One possibility to combine the phylogenetic information from all clusters is to build a "supertree", there are many ways to do that. In all, you first build a tree per orthologous group. Then you combine the trees into a consensus one, you can do this using the program duptree, which uses a gene tree parsimony approach finding the species tree that is imputes the lowest number of duplication in the family tree. It is quite powerful and enables the presence of paralogs in the trees (it actually exploits it).

The presence of paralogs in your tree hampers strategies such as gene concatenation or supertree approaches that do not enable duplications (e.g. the consensus approaches in PHYLIP).... a way out of this is to decompose the trees with duplication in trees containing only orthologs using the treeKO algorithm (http://treeko.cgenomics.org)

Hope it helps.

Toni

I can highly recommend the software MEGA4.

http://www.megasoftware.net/mega4/index.html

I found it really easy to set up, quite comprehensive and very approachable even for a non-expert in evolutionary biology.

There are two possible meanings of "consensus tree". One is when you compare hundreds of trees and plot the consensus tree as in bootstrapping phylogenetic analysis, and the other is to build a consensus sequence for groups of genes and

then build a phylogenetic tree from the resulting consensus sequences.

There are dozens of tools available for each of those routes. Almost all phylogenetics program packages such as PHYLIP, MEGA, DAMBE etc have tree bootstrapping and consensus built in. BioEdit and many other multiple sequence alignment editing

tools have consensus sequence building built in.

www.mbio.ncsu.edu/bioedit/bioedit.html‎

The treemaker and PhyML server at the HIV Databases, as well as most multiple sequence alignment servers, allow you to download the multiple sequence alignment in tree order so that the most similar sequences are next to each other in the alignment, making it easy to build a consensus for each group, whether you are

interested in one consensus for each homolog (alpha globin, beta globin etc

in many mammals) or one consensus for each organism (one alpha-beta-gamma

globin consensus for each mammal species).

http://www.hiv.lanl.gov/components/sequence/HIV/treemaker/treemaker.html

Since the maximum number of sequences per cluster is higher than the maximum number of species per cluster, i deduce your orthologous clusters contain (in)paralogs. I also assume the clusters are from different families and what you want to infer is the species tree. Let me know otherwise.

One possibility to combine the phylogenetic information from all clusters is to build a "supertree", there are many ways to do that. In all, you first build a tree per orthologous group. Then you combine the trees into a consensus one, you can do this using the program duptree, which uses a gene tree parsimony approach finding the species tree that is imputes the lowest number of duplication in the family tree. It is quite powerful and enables the presence of paralogs in the trees (it actually exploits it).

The presence of paralogs in your tree hampers strategies such as gene concatenation or supertree approaches that do not enable duplications (e.g. the consensus approaches in PHYLIP).... a way out of this is to decompose the trees with duplication in trees containing only orthologs using the treeKO algorithm (http://treeko.cgenomics.org)

Hope it helps.

Toni

Put all best trees in just one text file and import it in treefinder, in less than a minute you will get your consensus tree.

hope it helps

juan

Hi, there is the upper version of MEGA4 that you can easily use. If you are a new user I recommend you MEGA5 which is much easy than MEGA4.

Well, following the suggestion of Toni Gabaldón, if you have paralogs maybe some of the species doesn't have the complete set of genes, so my recommendation is to use supernetworks, with a similar meaning that supertrees. The weakness of supernetworks are the no existance, at this point, statistical support but you can compare with your supertree or individual trees. The strongest of this method is that you can visualize all species in one network and conflicting data.

To do this you can use the software SPLITSTREE 4, DENDROSCOPE, QUARTETNET or PHYLONET.

Good luck!

Peris

Hi, I use MEGA version 5.2.1. You can construct the tree you want using one of options under Phylogeny in the main window. I use this for my initial analysis. You can also use TreeView or FigTree after running a bayesian analysis.

use Mega version 5 for constructing the phylogenetic tree. it's easy to use.

Gudluck!

I recommend the Geneious software is easy to use and reliable

Mega vs 5 is very user friendly but for a new user you need to go thruogh the tutorials, Bio-edit is also very user friendly

Thanks Peris! Yes in some of my clusters I do not have complete set of orthologous genes. I will also try to check for supernetworks and compare them with supertree.

Mega 5 and phylip

Thanks you all for your suggestions. Thanks Toni for making my attention on (in)paralogs in my clusters. I identified that in some of my clusters I have 3-5 sequences of same species. Could you tell me how can I remove them in an automated way. I will also read about treeKO algorithm. My clusters are from different families but all of them functionally involve in immune system. I want to infer, how close my organisms are based on their immune proteins by the species tree.