Re: [modeller_usage] Modeller for alignment and modeling
To: Michael Meyer <>,
Subject: Re: [modeller_usage] Modeller for alignment and modeling
From: Modeller Caretaker <>
Date: Thu, 02 Oct 2014 16:56:28 -0700
On 10/2/14, 3:06 PM, Michael Meyer wrote:
Is there any way to
use the PIR format en route to automodel.make() to specify two chains
that aren't contiguous in a template?
If the chains are in the right order (e.g. A, C) you can include the
full sequence of chains A, B and C and then align the B chain with a set
of gaps in the model:
template AAAAAAAA/BBBBBBBB/CCCCCCCC*
model aaaaaaaa/--------/cccccccc*
If the chains are in the wrong order (e.g. C, A) then you can treat them
as two separate templates:
template1 CCCCCCCC/--------*
template2 --------/AAAAAAAA*
model cccccccc/aaaaaaaa*
Note however that this will discard any interactions (distance
restraints) between chains A and C.
Otherwise, your only option is to edit the PDB file and reorder the chains.
Furthermore, is there any way to use align2d to perform the two
sequence-two chain alignment in one step to create the appropriate
PIR file for multi-chain modeling. Right now, I'm doing two align2d
commands to create two separate PIR files, and then mashing them
together afterward.
You can certainly try to align both chains at the same time. align2d (as
long as you use a recent version of Modeller) will try to line up the
gaps, but since it's basically trying to globally align the two
sequences it could get confused if your chains are of very different
lengths or you do something crazy like try to align one chain in the
template with two in the model. Aligning the chains separately probably
works the best in this case.