Re: [modeller_usage] How to choose the best model?
To:
Subject: Re: [modeller_usage] How to choose the best model?
From: Charlotte Habegger-Polomat <>
Date: Wed, 07 Dec 2005 10:31:44 -0500
Cc:
Hello!
For my protein of interest, I generate 20 models with Modeller, use
Modeller's internal objective value to select the five best models, and
I evaluate the models with Procheck, Whatif, Vadar, Errat and Prosa2003.
I have not yet finalized my selection criteria after evaluation, but
I've checked that the selected models are all of satisfactory quality
according to these programs. These also help to pinpoint which areas of
my multiple alignment might be problematic.
For my research the parameter which i should consider for the best
model is from the evaluated program such as PROSA PROCHECK or
Verify3D. For Procheck , i use the model which give a quaility of
model within >80% core regions. That i used in my project so if any
recomments for my research please contact me.
Thank
Mr. JItrayut Jitonnom CSML
Chiang Mai University
Thailand
-----Original Message-----
From:
Sent: Fri, 02 Dec 2005 21:41:42 -0200
To:
Subject: [modeller_usage] How to choose the best model?
Dear Modeller users,
I'm generating diverses models to my protein (400 - is this to great?).
However I don't know how is the best parameter to look when choosing
the best
model. Objective function from Modeller, ProCheck parameters, Energy
parameters... What parameters you use and why? If using Ramachandran
plot as a
discriminative parameter, how great must be the parameters calculated
(for