Subject: Re: [modeller_usage] question about mutate_model
From: Modeller Caretaker <>
Date: Mon, 30 Jun 2008 09:03:09 -0700
Anna Marabotti wrote:
I would like to use the script mutate_model to evaluate different
conformations of the same mutant side chain inserted in a protein.
Therefore I used the script "mutate model" as suggested in the Wiki
section, and repeated the procedure 5 times, expecting to obtain 5
different conformations of the mutant side chain. Instead, I obtained
5 identical proteins with side chains perfectly superimposable
between each others. How can I manage myself in order to obtain 5
different models of the same mutation? Moreover, this is probably a
naive question, but since repeated MD cycles generally do not produce
identical results, why this protocol did not work in such a way?
If you start from the same initial conditions each time, your repeated
MD cycles had better produce identical results, otherwise your science
is not reproducable! Like most simulation packages, Modeller uses a
pseudorandom number generator, so will yield reproducable results if the
random seed is the same.
In your case, if you want to generate 5 different results from the
mutate_model script, modify it to use a different random seed for each
run. You can do this by changing the value of rand_seed when you create
the initial environ() object.