If the branch lenghth in species tree are necessary in my analysis using hyphy?

[Neo-xbx-00]

Dear hyphy users and developers,
I found it really convenient for using this tool to identity selections among species. Here, I face an issue and hope you to help me out. Now, I have only three species:
(1) I use Orthofinder software to get gene orthologs among these species, which include orthologs containing the single-copy genes in these species.
(2) Then, I choose the above single-copy orthologs and conduct analysis using hyphy. However, as I only have three species, I can not have branch length in my species tree, which looks like this, all branch lenth have been set to 1:
((Emal:1,Erad:1):1,Lbre:1):1;
So, I wonder if the species tree lack branch length will have negative influence on hyphy analysis and models?
Best regards!

[spond]

Dear @Neo-xbx-00,

If you provide branch lengths, HyPhy will use them as initial guesses. If you don't, HyPhy will makes its own guesses. In any case, these values are re-optimizied in the vast majority of analyses. For example (see the attached .zip for data files). It is better to provide a tree with no lengths than bad lengths.

1. Tree with reasonable lengths.

hyphy busted --alignment CD2.phylip --tree CD2-tree-with-lengths.nwk --starting-points 5
...

* Log(L) = -3414.47, AIC-c =  6910.74 (40 estimated parameters)
....
Likelihood ratio test for episodic diversifying positive selection, **p =   0.0031**.

2. Tree without lengths.

 hyphy busted --alignment CD2.phylip --tree CD2-tree-without-lengths.nwk --starting-points 5
...
* Log(L) = -3414.47, AIC-c =  6910.74 (40 estimated parameters)
....
Likelihood ratio test for episodic diversifying positive selection, **p =   0.0031**.

3. Tree with really wrong branch lengths (1000 each)

 hyphy busted --alignment CD2.phylip --tree CD2-tree-with-crazy-lengths.nwk --starting-points 5
...
* Log(L) = -3414.47, AIC-c =  6910.74 (40 estimated parameters)
....
Likelihood ratio test for episodic diversifying positive selection, **p =   0.0031**.

Best,
Sergei

CD2.zip

[Neo-xbx-00]

Just make some supplyment:
In my case, I test seven models; and I would like you to help me identify potential errors in the code:
For FUBAR model:
The code is: cat ../good_HOG.list | while read i; do hyphy fubar CPU=8 --alignment ../../2_pal2nal/pal2nal_results/${i}.pal2nal.fa --tree ./Litt_tree_all.txt --output ${i}.fubar.json > ${i}.fubar.log 2>&1; done
The tree is: ((Emal:1,Erad:1):1,Lbre:1):1;

For FEL model:
The code is: cat ../good_HOG.list | while read i; do hyphy fel CPU=8 --pvalue 0.05 --alignment ../../2_pal2nal/pal2nal_results/${i}.pal2nal.fa --tree ./Litt_tree_all.txt --output ${i}.fel.json > ${i}.fel.log 2>&1; done
The tree is: ((Emal:1,Erad:1):1,Lbre:1):1;

For MEME model:
The code is: cat ../good_HOG.list | while read i; do hyphy meme CPU=8 --pvalue 0.05 --branches test --alignment ../../2_pal2nal/pal2nal_results/${i}.pal2nal.fa --tree ../Litt_tree_E_foreground.txt --output ${i}.meme.json > ${i}.meme.log 2>&1; done
The tree is: ((Emal{test}:1,Erad{test}:1):1,Lbre:1):1;

For aBSREL model:
The code is : cat ../good_HOG.list | while read i; do hyphy absrel CPU=8 --branches test --alignment ../../2_pal2nal/pal2nal_results/${i}.pal2nal.fa --tree ../Litt_tree_Echino_foreground.txt --output ${i}.absrel.json > ${i}.absrel.log 2>&1; done
The tree is: ((Emal{test}:1,Erad{test}:1):1,Lbre:1):1;

And For Relax model:
The code is: cat ../good_HOG.list | while read i; do hyphy relax CPU=8 --test T --reference R --alignment ../../2_pal2nal/pal2nal_results/${i}.pal2nal.fa --tree ./Litt_tree_relax_model.txt --output ${i}.relax.json > ${i}.relax.log 2>&1; done
The tree is: ((Emal{T}:1,Erad{T}:1):1,Lbre{R}:1):1;

I'm worry about if I have some wrong understanding of hyphy analysis. Many Thanks !

[Neo-xbx-00]

Dear @Neo-xbx-00,

If you provide branch lengths, HyPhy will use them as initial guesses. If you don't, HyPhy will makes its own guesses. In any case, these values are re-optimizied in the vast majority of analyses. For example (see the attached .zip for data files). It is better to provide a tree with no lengths than bad lengths.

1. Tree with reasonable lengths.

hyphy busted --alignment CD2.phylip --tree CD2-tree-with-lengths.nwk --starting-points 5
...

* Log(L) = -3414.47, AIC-c =  6910.74 (40 estimated parameters)
....
Likelihood ratio test for episodic diversifying positive selection, **p =   0.0031**.

2. Tree without lengths.

 hyphy busted --alignment CD2.phylip --tree CD2-tree-without-lengths.nwk --starting-points 5
...
* Log(L) = -3414.47, AIC-c =  6910.74 (40 estimated parameters)
....
Likelihood ratio test for episodic diversifying positive selection, **p =   0.0031**.

3. Tree with really wrong branch lengths (1000 each)

 hyphy busted --alignment CD2.phylip --tree CD2-tree-with-crazy-lengths.nwk --starting-points 5
...
* Log(L) = -3414.47, AIC-c =  6910.74 (40 estimated parameters)
....
Likelihood ratio test for episodic diversifying positive selection, **p =   0.0031**.

Best, Sergei

CD2.zip

Well~ Many Thanks for your in time reply! I get it now.

[github-actions]

Stale issue message