Re: [modeller_usage] strategies for model refinement
To:
Subject: Re: [modeller_usage] strategies for model refinement
From: Thomas Evangelidis <>
Date: Sun, 9 Jan 2011 04:14:20 +0200
Cc: Irene Newhouse <>
On 7 January 2011 01:27, Mensur Dlakic <">> wrote:
Irene,
I know firsthand that Rosetta works even for proteins >200 residues,
meaning it will make them much better in terms of energy and
stereochemistry. Now, if you are expecting that the program will remove
absolutely all problems with the model so it gets a perfect Molprobity
score, that would be unrealistic. It seems to me that you already have a
pretty good model - it doesn't have to be free of all violations to be
useful. Even many legitimate protein structures, which are based on
experimental maps, have "issues" when analyzed by Molprobity.
That's very true!
Having said all of this, I have good experience with RAPPER when it comes
to loop modeling, assuming loops are of reasonable length. The program is here:
If you get the program to install and work properly, I can offer you a
small script that will take a PDB file and a range of residues for
refinement, and produce a new model resampled loops. Alternatively, you can
send me your PDB file and a range of residues to be refined and I will send
you the model back. If you are happy with it, then you can spend more time
installing and optimizing RAPPER.
Mensur, are you aware of any benchmark analysis that shows RAPPER's superiority on loop modeling over MODELLER? I am aware of one that compares Rosetta, MODELLER and CABS on loops up to 24 aa if I remember correctly, and shows that in short lengths all 3 programs are equivalent, but for longer loops the combination of MODELLER with CABS prevails.
I have several proteins with missing regions of varying length (8-47 aa) which I want to fold. I have tried PyRosetta and MODELLER in the past but I'm not impressed from the results of loop modeling, especially as the length increases. I guess these routines were designed to model flexible animo acid stretches that's why the predicted conformations adopt coiled coils. To this end I also tried I-TASSER server (which employs Monte Carlo with Replica Exchange for regions with no templates and is questionably more accurate in folding long aa stretches) by supplying my initial structure and excluding all homologues, but the meta-threading program (LOMETS) it implements always detects traces of "homology" to some irrelevant structures and uses them as templates. I guess my last resort is MD with Replica Exchange but I haven't found the time yet to set up my system and figure out a way to keep the rest of the model rigid, namely to fold only the missing regions.
This message is slightly off-topic, but I just wanted to share my experience (and desperation) with other people that might find it helpful (I hope not the desperation).