Question: How To Check The Reliability Of A Phylogenetic Tree
gravatar for kmkdesilva
6.8 years ago by
United States
kmkdesilva90 wrote:

Hi all,

I draw phylogenetic trees using MEGA. Other than the bootstrap value I could not find any way to check it's reliability.

Are there any other options in MEGA or some other software to calculate some statistical parameter and check reliability of inferred phylogeny. I'm new to this analysis and need your help.

Thank you in advance.

Regards, kalpi

statistics • 4.7k views
ADD COMMENTlink modified 6.8 years ago by DG7.1k • written 6.8 years ago by kmkdesilva90
gravatar for 5heikki
6.8 years ago by
5heikki9.0k wrote:

Do it with RAxML, do it with PhyloBayes. Are the trees the same?

ADD COMMENTlink written 6.8 years ago by 5heikki9.0k
gravatar for Spitshine
6.8 years ago by
Esch-sur-Alzette, Luxembourg
Spitshine640 wrote:

The bootstrap is not giving an answer to the reliability of a tree but the support for particular branches. It is prone to systematic errors, in particular long branch attraction just like any other measure.

If you want to assess the validity of your tree in the real world, you will have to learn the phylogeny of the species in consideration.

ADD COMMENTlink written 6.8 years ago by Spitshine640
gravatar for DG
6.8 years ago by
DG7.1k wrote:

It depends on what you mean by assess the reliability of the tree. It also highly depends on the type of phylogenetic analysis you are doing. Are you doing single-gene phylogenies? How many taxa are in your dataset? Multi-gene phylogenies? Are you interested in the phylogenetics for speciation or are you trying to infer something about the protein or gene in question? In general to get answers about support of your phylogenetic tree when trying to answer questions about the organisms in it you should:

1) Be doing multi-gene phylogenies. Single genes can have different evolutionary patterns than the species they are contained in for a variety of reasons. You should also be doing broad taxon sampling

2) Be doing Maximum-Likelihood or Bayesian phylogenetica, and not Neighbour-Joining. They are both more robust inferences

3) Model selection may be important for your particular gene(s) of interest. ModelTest is useful.

4) Ideally do more than one type of tree reconstruction (both Likelihood and Bayesian)

5) AIC tests or similar are also good ways to test how well your phylogeny holds up compared to an alternative reconstruction.

But again, the exact type of phylogenetic experiment you are doing, and the question you are trying to answer, are very important when considering how to validate how supported your phylogeny is.

ADD COMMENTlink written 6.8 years ago by DG7.1k
Please log in to add an answer.


Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.3.0
Traffic: 1224 users visited in the last hour