Creating consense tree using bootstrap value support not majority rule

Question

[Elephantdumbo]

Hey guys, I have few tree files in newnick format and i "cat" them into one big tree file . I want to constuct a consense tree using the bootstrap support not majority rule. I mean analyzing each tree alone i find correct relation in a clade that has like 100 bootsrap support or 95 % + but then they get lost when i use iqtree consense or phylip consense because the majority of the other tree has the wrong relation for this specific clade for example, so the question is How can i do consense based on support not majority ?

I hope i am clear enough. I am sorry I am not very familiar with phylogeny

Thank you so much

[Brice Sarver]

You need to clarify substantially and/or may be confused about what a consensus tree actually is. Are you concatenating Newick trees and treating them as bootstrap replicates, or have you actually performed bootstrap replicates? Where does this come into play? I can provide suggestions in a comment but can't come up with an answer until the inputs and goals are clearer.

Alternatively, if you want to take a list of trees and return a tree where the only nodes present are those that are in 100% of your trees, you can just increase the threshold from 50 (present in the majority) to 100 (present in all, i.e., a strict consensus tree). Remember, bootstrap support is the presence of that bifurcation that is itself underpinned by the original data. Alternatively, you could collapse nodes where your support is below a given threshold (using, say, R) on each tree and then do the suggestion above.

To start, here's how to do it in PAUP* and how to do it with RAxML.

[Elephantdumbo]

Thanks your reply. The idea is i am getting different phylip files belong to different reference sequences, but they contain the same samples. I want to infer phylogentic relation ship based on these Philip files. What do you think make sense ?

I want to do something the bootstrap thing you said, but i keep messing it up in r. Can you explain further how to do it, so i can do the majority rule ?

[Brice Sarver]

By different reference sequences, I'm assuming you mean different loci but the same individuals? This is a perfect candidate for species tree approaches, which attempt to account for discordance among loci due to a number of factors (depending on the approach) and estimates the species tree given a set of gene trees (that were previously estimated). If this is what you want to do, I can point you in the right direction with a formal answer to your question.

[Elephantdumbo]

Yes please. This is exactly what i want

Answer 1

[Brice Sarver]

From discussion in the comments above:

For a rapid, quartet-based approach, check out ASTRAL. For a fully Bayesian implementation, check out *BEAST or SNAPP, implemented in BEAST2, depending on how your data are structured. You can also check out BPP.

There are other approaches that may be useful depending on your data and the evolutionary forces at play in your system, but these should be a good starting point for further investigation.

Hope this helps!