Re: [modeller_usage] scoring of the full loop-models
To: malkeet singh Bahia <>,
Subject: Re: [modeller_usage] scoring of the full loop-models
From: Modeller Caretaker <>
Date: Tue, 16 Apr 2019 10:17:29 -0700
On 4/16/19 4:28 AM, malkeet singh Bahia wrote:
In the loop refinement process, is the DOPE score reported only for the
refined region or full protein model?
Neither. It is the score for the refined region interacting with the
entire model. DOPE is a pairwise atomistic score, so it is evaluated for
all nearby pairs of atoms where at least one atom is in the refined region.
If only for the refined region, is
there any 'python code', which could be clubbed with loop refinement
script to report the DOPE scores of the complete protein models as well?
This shouldn't be necessary, since DOPE scores are only reliable for
ranking, and since the non-refined parts of the system don't move, the
contributions to the score from the non-refined-non-refined atom pairs
should be constant, so the rankings won't change. But sure, you can call
assess_dope directly on whatever selection you like:
If you want to tie this in to loopmodel automatically, you can write a
little function to do it and assign that function to loop_assess_methods
in the usual fashion. See modlib/modeller/automodel/assess.py for the
existing (very simple) functions.