Phylogenys with multi genes

[joaomoreiraoliveira]

Hi guys, I read the documentation and it isn't clear to me if it's possible to do a phylogeny with more than 1 gene and diferents evolutive models being used for each gene to infer a single tree.

Currently, my workaround is to infer the phylogenys separately and then build a concensus tree, but I'm note sure if is the best way to do this

Thank you for the help!

[GavinHuttley]

Hi @joaomoreiraoliveira,

The best solution depends on the nature of your data. Do you have exactly the same species for each gene?

[joaomoreiraoliveira]

I had exactly the same species for each gene but not all the samples had the both genes. Some of then had just COI, but in general, the samples had COI and 16S

[GavinHuttley]

So your design is samples x genes x species. This means for each gene, the tree topologies will have the exact same tip labels. That means a consensus tree is appropriate.

In contrast, a supertree method (e.g. sc-supertree) is appropriate when the genes have different species.

Hope that's useful!

[joaomoreiraoliveira]

I'm having a problem with the output, the output of consensus tree don't bring me the bootstrap values. Here in my code:

tree1 = build_tree(coi, 'HKY+F+I', bootstrap_replicates=1000, num_threads=4)
tree2 = build_tree(sixts, 'GTR+F+G4', bootstrap_replicates=1000, num_threads=4)
tree_with_likelihood = fit_tree(coi, tree1, model='HKY+F+I', bl_fixed=True)
log_likelihood = tree_with_likelihood.params["lnL"]
tree = consensus_tree([tree1, tree2])
tree.write('concatenada.nwk')

[GavinHuttley]

Hi @joaomoreiraoliveira , I'm following up on this with another developer. Feel free to ask for an update if I haven't replied by mid next week.

[GavinHuttley]

Hi @@joaomoreiraoliveira,

thanks to @thomaskf , we've established there's a bug here so I'm converting this into a bug report. We will get a fix out soon.

[GavinHuttley]

@joaomoreiraoliveira , we have released a new version of piqtree which fixes this. To upgrade do

pip install piqtree==0.8.3

[thomaskf]

Hi, thanks for the question!

First, we have been working on the issue you encountered in your workaround solution.

On the other hand, please note that a standard approach to your question is to use a partition model which can estimate a single tree jointly while allowing each gene its own substitution model and rates. Right now, piqtree does not support partitioned analysis, while IQ-TREE 3 can handle it with a partition file (-p/-q/-Q plus a NEXUS file). You may refer at the section "Ultrafast bootstrapping with partition model" on IQTREE documentation (https://iqtree.github.io/doc/Advanced-Tutorial)