How To Set The Substitution Model In Phyml?
1
2
Entering edit mode
12.5 years ago
Dejian ★ 1.3k

I am drawing two phylogenetic trees. To select the most approapriate model, I ran ProtTest with the aligned amino acid sequences. The ProtTest suggested that the best models for the two genes are Blosum62+G and WAG+I+G. However, When I try to build the trees using PhyML, I do not know how to set the parameters to fit to the suggested models. The option -m (or --model) can accept Blosum62 and WAG. But, which options can accept G and I in the suggested models?

• 8.3k views
ADD COMMENT
5
Entering edit mode
12.5 years ago
Leo Martins ▴ 230

If I remember correctly:

-d aa -c 4 -a e -m Blosum62
-d aa -c 4 -a e -m WAG -v e

for Blosum62+G and WAG+I+G, respectively.

The option -c is for the number of discretized categories for the gamma rates (G), and you can choose a different (larger) value; 4 is the default. The option -a is for the gamma shape parameter (G) and -v is for the fraction of invariant sites (I), and both accept the "e" value which means that we want to estimate these values (you could also force them to fixed values).

I don't know if it's relevant for you, but ProtTest3 can also give you the ML tree of the best model (since it uses phyML behind the scenes). The option is -t1 for tree output in newick format and -t2 for text format. E.g.

java -jar prottest-3.0.jar -i align.phy  -t1  -all-matrices -all-distributions -F -S 2 -o prott.out
ADD COMMENT
0
Entering edit mode

A great help, Leo. Many thanks. I tried ProtTest3. When trying to get a consensus tree, I set -tc 0.5. I thought the option (-tc) was responsible for the construction of a consensus tree. I got a tree with all the support value of 100. I am a bit surprised because I never got that high support value with these sequences before. Is that the right way to build a consensus tree by bootstrapping in ProtTest3? Furthermore, I did not find the option for the times of bootstrapping.

ADD REPLY
0
Entering edit mode

There is another thread talking about phyml consensus trees. It seems that it needs extra effort to get a consensus tree using PhyML. Is it the same when using ProtTest3? Many thanks! Building Consensus Trees Of Phyml Bootstraps

ADD REPLY
0
Entering edit mode

This is the command I use to build the consensus tree:

java -jar prottest-3.0.jar -i trimal.phy -o trimal.phy.prott.phyml -S 2 -t1 -threads 4 -all-matrices -all-distributions -tc 0.5
ADD REPLY
0
Entering edit mode

There is another thread talking about phyml consensus trees. It seems that it needs extra effort to get a consensus tree using PhyML. Is it the same when using ProtTest3? Many thanks! Building Consensus Trees Of Phyml Bootstraps

ADD REPLY
0
Entering edit mode

I have never used this option, but I'm almost sure that the "consensus" they refer to is not a bootstrap consensus, it is the consensus among all models.

hat is, for each model ProtTest uses phyML once to find the maximum likelihood (ML) tree and ML parameters. And then the option -tc will use the ML trees (one per model) to calculate the consensus tree. If for all models the ML tree found was the same (maybe with different branch lengths), then you'll have 100% support.

ADD REPLY

Login before adding your answer.

Traffic: 2116 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6