Dear Jacob,
What do we gain? As Dale pointed out, we are already abusing either occupancy, B-factor or delete the side chain to compensate for our inability to tell the user that the side chain is disordered. With your proposal, we would fudge both occupancy and B-factor, which in my eyes is even worse as fudging just one of the two.
Also, who should decide on the magic number: the all-knowing gurus at the protein data bank? Maybe we should really start using cif files, which allow to specify coordinate uncertainties.
Best regards,
Herman
-----Original Message-----
From: Jacob Keller [mailto:[log in to unmask]]
Sent: Thursday, March 31, 2011 4:43 PM
To: Schreuder, Herman R&D/DE
Cc: [log in to unmask]
Subject: Re: [ccp4bb] what to do with disordered side chains
Why not have the "b-factors take care of it" until some magic cutoff number? When they reach the cutoff, two things happen:
1. Occupancies are set to zero for those side chains, to represent our lack of ability to model the region,
2. B-factors are set to exactly 500, as a "flag" allowing casual b-factor-savvy users to identify suspicious regions, since they will probably not see occupancies, but *will* see b-factors. Therefore, all 0-occupancy atoms will automatically have b-factors = 500. I believe it is true that if the occupancies are zero, the b-factors are totally irrelevant for all calculations?
Doesn't this satisfy both parties?
Jacob
On Thu, Mar 31, 2011 at 9:22 AM, <[log in to unmask]> wrote:
> Dear Quyen,
> I am afraid you won't get any better answers than you got so far.
> There is no holy bible telling you what to do with disordered side
> chains. I fully agree with James that you should try to get the best
> possible model, which best explains your data and that will be your decision. Here are my 2 cents:
>
> -If you see alternative positions, you have to build them.
> -If you do not see alternative positions, I would not replace one
> fantasy (some call it most likely) orientation with 2 or 3 fantasy orientations.
> -I personally belong to the "let the B-factors take care of it" camp,
> but that is my personal opinion. Leaving side chains out could lead to
> misinterpretations by slightly less savy users of our data, especially
> when charge distributions are being studied. Besides, we know (almost)
> for sure that the side chain is there, it is only disordered and as we
> just learned, even slightly less savy users know what flaming red side
> chains mean. Even if they may not be mathematically entirely correct,
> huge B-factors clearly indicate that there is disorder involved.
> -I would not let occupancies take up the slack since even very savy
> users have never heard of them and again, the side chain is fully
> occupied, only disordered. Of course if you build alternate positions,
> you have to divede the occupancies amongst them.
>
> Best,
> Herman
>
> ________________________________
> From: CCP4 bulletin board [mailto:[log in to unmask]] On Behalf Of
> Quyen Hoang
> Sent: Thursday, March 31, 2011 3:55 PM
> To: [log in to unmask]
> Subject: Re: [ccp4bb] what to do with disordered side chains
>
> We are getting off topic a little bit.
> Original topic: is it better to not build disordered sidechains or
> build them and let B-factors take care of it?
> Ed's poll got almost a 50:50 split.
> Question still unanswered.
> Second topic introduced by Pavel: "Your B-factors are valid within a
> harmonic (small) approximation of atomic vibrations. Larger scale
> motions you are talking about go beyond the harmonic approximation,
> and using the B-factor to model them is abusing the corresponding mathematical model."
> And that these large scale motions (disorders) are better represented
> by "alternative conformations and associated with them occupancies".
> My question is, how many people here do this?
> If you're currently doing what Pavel suggested here, how do you decide
> where to keep the upper limit of B-factors and what the occupancies
> are for each atom (data with resolution of 2.0A or worse)? I mean, do
> you cap the B-factor at a reasonable number to represent natural
> atomic vibrations (which is very small as Pavel pointed out) and then
> let the occupancies pick up the slack? More importantly, what is your reason for doing this?
> Cheers and thanks for your contribution, Quyen
>
> On Mar 30, 2011, at 5:20 PM, Pavel Afonine wrote:
>
> Mark,
> alternative conformations and associated with them occupancies are to
> describe the larger scale disorder (the one that goes beyond the
> B-factor's capability to cope with).
> Multi-model PDB files is another option.
> Best,
> Pavel.
>
>
> On Wed, Mar 30, 2011 at 2:15 PM, VAN RAAIJ , MARK JOHAN
> <[log in to unmask]> wrote:
>>
>> yet, apart from (and additionally to) modelling two conformations of
>> the side-chain, the B-factor is the only tool we have (now).
>> Quoting Pavel Afonine:
>>
>> > Hi Quyen,
>> >
>> >
>> > (...) And if B-factor is an estimate of thermo-motion (or static
>> > disorder),
>> >> then would it not be reasonable to accept that building the
>> >> side-chain and let B-factor sky rocket might reflect reality more
>> >> so than not building it?
>> >>
>> >
>> > NO. Your B-factors are valid within a harmonic (small)
>> > approximation of atomic vibrations. Larger scale motions you are
>> > talking about go beyond the harmonic approximation, and using the
>> > B-factor to model them is abusing the corresponding mathematical
>> > model.
>> > http://www.phenix-online.org/newsletter/CCN_2010_07.pdf
>> >
>> > Pavel.
>> >
>>
>> Mark J van Raaij
>> Laboratorio M-4
>> Dpto de Estructura de Macromoléculas
>> Centro Nacional de Biotecnología - CSIC c/Darwin 3, Campus
>> Cantoblanco
>> 28049 Madrid
>> tel. 91 585 4616
>> email: [log in to unmask]
>
>
>
--
*******************************************
Jacob Pearson Keller
Northwestern University
Medical Scientist Training Program
cel: 773.608.9185
email: [log in to unmask]
*******************************************
|