Re: [modeller_usage] Two domains get oriented randomly although modelled from a single template
To: Чугунов Антон <>
Subject: Re: [modeller_usage] Two domains get oriented randomly although modelled from a single template
From: Modeller Caretaker <>
Date: Thu, 30 Jun 2011 08:03:58 -0700
Cc:
On 6/30/11 3:01 AM, Чугунов Антон wrote:
I'm building a model of a voltage-sensitive ion channel using Kv1.2/2.1
paddle chimera structure as a template (pdb: 2r9r). Although
voltage-sensing (S1-S4) and pore (S5-S6) domains of the template come as
a single chain, I get in the model more or less arbitrarily oriented VS
and pore domains
This shouldn't happen when building a model using a single template,
since Modeller restrains internal CA-CA distances in the model to their
values in the template. However, for efficiency there is a cutoff on
this distance - by default automodel only restrains CA-CA distances that
are less than 14 angstroms in the template. This might not be long
enough to include enough CA-CA restraints that span the two domains.
Thus, the hinge between the two domains will only be constrained by
stereochemistry, and will likely be quite flexible.
If you really want the angle to match that in the template, there are
two things you can try:
1. Add one or more user-defined distance restraints between the two domains.
2. Increase automodel.max_ca_ca_distance from 14.0 to a value large
enough to include some CA-CA pairs that span the two domains.