How to derive phylo tree from two genes

Question

[shadowstep]

I did multiple align for two genes, and got two multiple alignment files, each has nine sequences of the same nine species, for simplicity, I illustrate an example with 3 sequences.

Alignent 1 :

X    AGCCTC

Y    ACCCTC

Z    ACGGTC

Alignent 2:

X    CGGCATCC

Y    CCGCATAC

Z    AGCCATCC

Now I need to build a phylo tree from the two genes. (In fact they are rbcL and matK from chloroplast )

1) Is it correct to join them into one big alignment and estimate a plylo tree?

2) If 1) is no correct, is it possible to co-estimate a phylo tree from two alignments?

Thanks for your help

[lieven.sterck]

Not sure about mega but there are others that can do progressive alignment (eg. mafft, muscle, ... from the top of my head) . That being said: normally you build phylogentic trees per gene family (and thus from one alignment file). I don't think that joining 2 distinct gene families in one tree makes much sense

[lieven.sterck]

are you looking to obtain a phylogenetic tree to illustrate the taxonomic position (== species tree) of your species of interest or are you interested in the phylogenetic relationships within (between) those 2 gene families (== gene tree) ?

In case of 1 : what you need is a concatenate alignment of your sequences. in that case you need to 'paste' both sequences of species X together and to the same for species Y and so on . based on that MSA you can then calculate the species tree

[shadowstep]

Hi lieven.sterck, I'm going to calculate phylogenetic diversity of a plant community, maybe what I need is a species tree. Is there any command line tool available?

[Joe]

You can compute a species/consensus tree with the program ASTRAL-II, after generating gene trees with a tool of choice (I'd suggest IQ-Tree or similar).

Its not going to be much of a consensus from just 2 different genes though, and 9 representatives may not produce the most robust trees.

You could use something like ete3 compare to measure how similar they are. If the ancestry of those genes is the same, your consensus tree will be == to the topology of either/both of the gene trees.

Another option might be to do concatenated alignment trees, but I don't think this would be particularly appropriate.

[shadowstep]

Thank you, one more question, when calculating phylo tree through maximum likelihood estimation, can I summarize a "best" tree from 100 bootstraps?

[lieven.sterck]

yes, though 100 would be kind of a minimum (more is better here)

[shadowstep]

1000 is too complicated for my computer though. I calculated 100 bootstraps using RAxML command line, but I don't know how to get this best tree, I searched a lot but found nothing

[Joe]

If you follow the RAxML documentation, there should be an output file called 'bestTree" or something to that effect. I would try for at least 300 bootstraps if you can manage it on your machine.

There is a second step in the process to add the bootstraps on to the tree nodes from the bootstrap trees file.

In short, follow the manual (it's only 3 or 4 steps).

[shadowstep]

I think I've got the point, if I don't do bootstrap, then the auto-generated best tree is actually the best, and indeed bootstrapping is intended to gauge nodal support, am I right?

[Joe]

Theoretically yes, but the 'best' tree, may have polytomies, and you wont know that unless you look at the bootstraps. A tree with no bootstraps at all is pretty much meaningless.