Empirical data test

[Hao Zhang]

Dear  GeneRax community,

May I ask how to compare the joint ML scores of the gene family trees (GFTs) obtained from other tools? like fig. 9 in the paper https://doi.org/10.1093/molbev/msaa141.

Is it achieved by using the GFT from other tools as the initial  tree in GeneRax? 

Thanks in advance.

Hao

[Benoit Morel]

Dear Hao,

All you need is to disable the GFT topology optimization, e.g. with

--strategy EVAL

The total likelihood should be outputed in the log files, and the per-tree likelihoods in the stats.txt files (see https://github.com/BenoitMorel/GeneRax/wiki/GeneRax#outputs)

Let me know if that helps!

Benoit

[Hao Zhang]

Dear Benoit,

Thanks for your kind help. I tried to provide the flag "--strategy EVAL", but the job finished very soon and I got the results with JointLL, LibpllLL and RecLL likelihoods equal to 0. 

The stdout has been attached for your reference. I noticed that the "unrooted reconciliation likelihood" was set to "OFF" by default, how can I open it?

BTW, gene tree and species tree used in this test are unrooted. 

Many thanks and best regards.

====

[00:00:00] GeneRax 2.0.1
Logs will also be printed into GeneRax/generax.log
GeneRax was called as follow:
generax -f families_ALE -s genome.tree.mo --strategy EVAL

General information:
- Output prefix: GeneRax
- Families information: families_ALE
- Species tree: genome.tree.mo
- MPI Ranks: 1
- Random seed: 123
- Reconciliation model: UndatedDTL
- DTL rates: global rates
- Infer ML reconciliation: ON
- Unrooted reconciliation likelihood: OFF
- Prune species tree mode: OFF

Gene tree correction information:
- Gene tree strategy: EVAL
- Max gene SPR radius: 0

[00:00:00] Filtering invalid families...

[00:00:00] Starting species tree initialization...
[00:00:00] End of species tree initialization
[00:00:00] Filtering invalid families based on the starting species tree...

End of instance initialization
[00:00:00] Gathering statistics about the families...
[00:00:00] Input data information:
- Number of gene families: 5
- Number of species: 62
- Total number of genes: 113
- Average number of genes per family: 22
- Maximum number of genes per family: 23
- Species covered with the smallest family coverage: "GNM9" (covered by 0/5 families)
- Average (over species) species family coverage: 1

Free parameters 3
Rates : (0.2, 0.2, 0.2, score = 0)[00:00:00] Reconciling gene trees with the species tree...
[00:00:00] Terminating the instance..

[00:00:00] DTL rates: D=0.2 L= 0.2 T=0.2
[00:00:00] Reconciliation likelihood: 0
[00:00:00] Results directory: GeneRax
[00:00:00] End of GeneRax execution

====

Hao

[Benoit Morel]

Dear Hao,

This looks like a bug that we fixed in a more recent version of GeneRax (the current version is 2.0.4). Could you please try to update GeneRax?

If you are using BioConda and that you are using the latest package, let me know, because maybe the package is not up to date with github. If so, can you try to install with github to see if it fixes the issue?

Please ignore the "- Unrooted reconciliation likelihood: OFF", the name is a bit misleading. Both modes assume the following:

- the input species tree is rooted

- the input gene trees are unrooted

Which brings me to the following: you need a rooted species tree. With your command, GeneRax will arbitrarily root the species tree, which is not very desirable. With the appropriate command line, GeneRax could try to estimate the root of the species tree, but I would not recommend it if you only have 5 input families.

Best,

Benoit