You are here

ddG Monomer Low Res Protocol Convergence

3 posts / 0 new
Last post
ddG Monomer Low Res Protocol Convergence

Dear Developers,

First question!! :-

I am using DDG Monomer to calculate ddgs for some mutants. Following the paper Kellogg-2011 I use both the Row3 (Low Resolution) and Row16 (High Resolution) protocols.

Also as is recommended in the manual page for ddg monomer ( I iterate 50 times for both protocols.

While I get a fair amount or variation between the calculated energies for the High Res protocol, I get (50) identical energy for each iteration for the Low Res protocol. (One would expect the High Res protocol to give different energies at each variation since there is backbone movement along with side chain repacking.)

Is this because I am doing something wrong or is it that the Low Res protocol converges quickly to a single value as only side chains in 8 Ang radius are being repacked?
The flag file I use for High Res is identical to the one given in manual page.

For the Low Res Protocol, I use the following flags:

-in:file:s min_cst_0.5.xxxx_0001.pdb
-ddg:mut_file mutantFile.inp
-ddg:weight_file soft_rep_design
-database /opt/ROSETTA/v3.3/rosetta_database/
#-fa_max_dis 9.0
-ddg:iterations 5
-ddg:local_opt_only true
-ddg::opt_radius 8.0
-ddg::min_cst false
-ddg::ramp_repulsive false
-ddg::sc_min_only false
-ddg::mean false
-ddg:min true
-ddg:dump_pdbs true
-ddg:suppress_checkpointing true
#-mute all
-ddg:output_silent true

These are all the flags given in the manual page PLUS a few additional one I took from the supplementary info (which were there for the older "ddg with fixed bb" program) of the same paper. Even when I take just the flags given in the man page, I get exactly identical values.

Second Question!! :-

Just to check if whatever I was doing was right or wrong, I thought of running this on a small part of the author's dataset i.e. the ProTherm Database. However the ProTherm database gives only the ddg values where ddG is defined as Mutant - WildType and dG is defined as "dG of Unfolding"

I think DDG Monomer also calculates ddG as Mutant - WildType (right??) however does Rosetta calculate dG of Folding or dG of Unfolding? Main question being - How did the original paper compare the calculated ddG values with those given in ProTherm database? (so that I can do the same...)

Also is it possible for anyone to email me the calculated ddG values of some 10-20 mutants of any protein so that I can compare and say for sure that whatever I am doing is exact and correct. (Or should I mail this request to the author of the paper?)

Third Question!! :-

I am having some trouble installing the mpi enabled version of rosetta. However reading through the forums, I realised that most of rosetta is not written to be mpi enabled (ddG Monomer isnt for one which I read in another post). So then is there any obvious advantage of installing with "extras=mpi" or should I just do a normal install and submit multiple jobs in different folders?

I apologize if things are getting a bit too cramped here but I was not sure if the smaller questions warranted their own threads...


p.s. : I have applied for my own account to post in this forum but the one time login link always turns out to be invalid/already-used. I have mailed to support regarding this. But while they replied, I thought I would post using Harsh's account who is a friend of mine (Different department, same university. Thanks Harsh!) Apologies if am doing something illegit!

Post Situation: 
Thu, 2012-03-01 00:45

A) and B); you need to talk to Liz Kellogg. Feel free to email directly; I'll forward this along to her.

C) MPI will make it harder to run code that is not MPI enabled, because you'll need to use mpirun even though you have separate jobs. Don't compile MPI unless you intend to use it. Submit in separate folders.

Thu, 2012-03-01 04:41

Hi Steven,

Thanks for your quick reply. Will email directly then. Thanks for forwarding this as well.


Thu, 2012-03-01 09:03