Hi All,

I am trying to test if selection is enhanced in 2 species relative to the rest of the tree (about 20 other species), where each of these 2 species can have different distributions for omega.

The data set consistes of sequence data from a nuclear gene, between one and 3 individuals per species representative.  The alleles from these individuals do not necessarily form monophyletic groups.. i.e. there is incomplete lineage sorting/transspecies polymorphism.  

How can I test the hypothesis that selection is enhanced in these 2 lineages, with omega varying between them?

Hi Matt,

The easiest exploratory analysis is to run your data through GA Branch on Datamonkey.
The method will estimate MEAN dN/dS for each branch in the tree; you can then look post processing results ("Get a [.csv table] of model averaged probabilities that any two branches have the same dN/dS.") to see if any two branches have the SAME dN/dS. The nice aspect of this method is that it does not force the "background" lineages into a uniform selection regime.

Alternatively, you can run a model which assigns three different dN/dS distributions to branch classes in your tree and compare them.
For example, you can test the hypothesis that the means of each of the lineages is greater than the background.

We can discuss this in more detail, but the relevant (simple) tests are described in Multimedia File Viewing and Clickable Links are available for Registered Members only!!  You need to Login .

Sergei

Can you do GA branch on HYPHY, my alignment is too big for GA Branch on Datamonkey?

Matt

Hi Matt,

How many sequences do you have? You can do GA Branch in HyPhy, but you need an MPI cluster for it to finish in a reasonable amount of time.

Sergei

Hi All,

I have one dataset with 121 sequences, and another with 78. They are short however, ~300bp each.

Also, I have access to a very large cluster, i.e. thousands of compute nodes, that I usually use for Illumina work, but I suppose I could put HYPHY on there.  Up until now, HYPHY has been running for me on a small 16 node cluster.

I'm still not quite sure how to run GABranch on HYPHY, it isn't mentioned by name, either in the .bf files or in the menu options.  

Thanks

Hi Matt,

The files for running GA Branch analyses locally can be found here: Multimedia File Viewing and Clickable Links are available for Registered Members only!!  You need to Login
Let me know if you have issues...

Sergei

Hi All,

Issues with the GABranch- or more likely the MPI install..
[code]macmanes@macmanes:~/hyphy/HYPHY$ /usr/lib64/mpich/bin/mpirun -np 2 ./HYPHYMPI BASEPATH=/home/macmanes/hyphy/HYPHY   /home/macmanes/hyphy/GABranchFiles/ModelSelectorBranchLocal.bf
Fatal error in MPI_Comm_rank: Invalid communicator, error stack:
MPI_Comm_rank(106): MPI_Comm_rank(comm=0x5b, rank=0x7fffe6724f9c) failed
MPI_Comm_rank(64).: Invalid communicator
macmanes@macmanes:~/hyphy/HYPHY$[/code]

In fact, looking at things, I can't get HYPHYMPI to run anything.. I checked out the latest version from SVN this AM.

The install seemed to go fine- no errors or anything. I am using MPICH, as I can't seem to get it to compile with OpenMPI.

Hi Matt,

Haven't seen this error before -- not sure what it is about. Looks like a generic MPI error, not something HYPHY specific.
Try with openMPI as well, by replacing


Code (]

COMPILER="g++";
COMPILERC="gcc";

[/code):

with COMPILER="g++"; COMPILERC="gcc"; [code] COMPILER="mpic++"; COMPILERC="mpicc";

in build.sh.

Sergei