Rooted Tree Input for Ancestral Reconstruction

[dafnago2]

Dear IQ-TREE team,
I would appreciate it if you could advise me on how to perform ancestral sequence reconstruction using a rooted tree as input.
Thank you!

[trongnhanuit]

Hi @dafnago2,
Thank you for using IQ-TREE.
To reconstruct ancestral sequences using a rooted tree, you can try the following command.

./iqtree3 -s <aln.phy> -te <tree.nwk> -m <nonreversible_model> -asr

where

• -s <aln.phy> specifies an input alignment.
• -te <tree.nwk> specifies an input (rooted) tree and keeps the topology fixed.
• -m <nonreversible_model> specifies a model. Here, you should use a non-reversible model to avoid IQ-TREE from unrooting your rooted tree. For DNA, you can consider UNREST (but be aware of overparameterization) or Lie Markov models; for protein, you can consider NQ.bird, NQ.insect, etc (see Substitution Models).
• -asr activates ancestral sequence reconstruction.

You may also consider adding -T <num_threads> to speed up your analysis using multiple threads/cpus.

Hope it helps. Please feel free to reach out to us if you have any questions. Thanks.

Nhan

[dafnago2]

Hi Nhan,

Hi, thanks for your earlier reply and the suggested command line!
I’d like to clarify my original question. I ran the command using a rooted phylogenetic tree as input, but in the output, the original root placement from my input tree was not preserved.
Is there a way to ensure that the location of the root in the input tree is retained in the output? Or alternatively, is there a flag or option I should use to explicitly preserve the original root placement?

Thanks again for your help!
Dafna

[trongnhanuit]

Hi Dafna,
Could you double-check whether you used -te (not -t) to specify your rooted tree and -m to specify a non-reversible model (e.g., UNREST, Lie Markov models for DNA or NQ. for protein data) in your test?
With these two inputs, IQ-TREE still keeps the root position in my tests.
If you did specify these inputs and still encountered the issue, could you please share with us an example data and the command you ran? Thank you.

Nhan

[dafnago2]

Hi Nhan,
If there is no reason to assume a non-reversible model, is there a way to keep the root position when specifying a reversible model?
Thank you!
Dafna

[trongnhanuit]

Hi Dafna,
When a reversible model is used, IQ-TREE automatically unroots the tree. Out of curiosity, would it be possible to instead use ASR at the internal node closest to the original root position for your research?

Hi @bqminh,
Is there any hidden option in IQ-TREE that allows users to preserve the root position when providing a rooted tree with a reversible model? Thanks.

Nhan

[dafnago2]

Dear Nhan and the IQ-TREE team,
I ran IQ-TREE with a reversible model. The output tree is rooted, but not at the expected position. To compare child nodes with their parents, I need the correct root placement.
As a workaround, I rerooted the output tree using the known outgroup, BAL. However, the true root (the parent of BAL) has no corresponding sequence from IQ-TREE, so I am unable to make any comparison to the root sequence.
I have attached the following files:

1. Input tree: YDR157W.tree.input.4iqtree.txt
2. Input MSA: YDR157W.MSA.input.4iqtree.txt
3. IQ-TREE output tree: YDR157W.MSA.input.4iqtree.treefile.txt
4. Rerooted tree (BAL as outgroup): YDR157W.MSA.input.4iqtree.treefile.rerooted.txt

This is the command I ran:
./iqtree3 -s <MSA.input.file> -te <tree.input.file> -m GTR -asr

How can I compare to the root sequence in this case?
I would greatly appreciate your help in resolving this issue.
Best regards,
Dafna

YDR157W.MSA.input.4iqtree.treefile.rerooted.txt
YDR157W.MSA.input.4iqtree.treefile.txt
YDR157W.MSA.input.4iqtree.txt
YDR157W.tree.input.4iqtree.txt

[trongnhanuit]

Hi Dafna,

I've checked and found that the IQ-TREE output tree (YDR157W.MSA.input.4iqtree.treefile.txt) is unrooted (not rooted, which is expected since you used a reversible model). By our definition, a rooted tree has a top node with exactly two children, whereas an unrooted tree has the top node with three children.

When a reversible model is used, IQ-TREE automatically unroots the input tree and assigns the top node to the one connected to the first taxon in the alignment. So, I moved the “BAL” sequence to the beginning of the alignment, and the resulting top node in the output unrooted tree was close to your original root position.

I've attached the reordered alignment - could you please try rerunning the command with this alignment and see if the top node aligns closely enough with your expected root?

YDR157W.MSA.input.4iqtree_reordered.txt

Cheers,
Nhan