I don't know anything about the practicalities of PDB2PQR but it
would seem to me that you have to calculate a potential for the molecule
with each conformation.  Then you would say "This is the potential with
the A altloc and THIS is the potential with B".  There will be no
individual molecule with the average potential so the average has no
chemical meaning.

   Of course life gets even harder when you have multiple side chains
with multiple conformations.  The combinatorials grow quickly.  If you
don't believe that the conformations are all tied to each other you have
to say that the ensemble of conformations leads to an ensemble of
potentials.  Somehow your chemistry has to work in the presence all this
variability and THAT is the interesting question you have to answer.

   I don't know what an ensemble of potentials looks like so one would
have to calculate some and see what their properties are.  I'm not aware
that anyone has done this, but my literature search has been very limited.

Dale Tronrud

On 12/2/2017 5:51 AM, Sam Tang wrote:
> To add to the discussion, could I raise a relevant question about
> generating ESP (Apologies to Jiri if this distracts too much from your
> initial thread). 
> 
> In our structure in hand, the density for two conformations of the side
> chain are clearly seen and they could be modeled. This brings a bit of
> problem because the positive charge becomes more prominent with two
> conformations there than with one. So what do we usually do when
> generating ESP for such structures with alternate conformations? Do we
> remove one before the calculation?
> 
> PS - I use online PDB2PQR server to do my calculation with PARSE field.
> I did notice from some old archived discussion on the Web that it
> ignores one conformation by default. But this seemingly is not the case
> in newer versions?
> 
> Regards
> 
> Sam
> 
> School of Life Sciences, CUHK
> 
> On 2 December 2017 at 02:59, Robbie Joosten <[log in to unmask]
> <mailto:[log in to unmask]>> wrote:
> 
>     If you cannot trust the surface of your protein, perhaps you should
>     not look at the the potential on the surface. Instead you can look
>     at the field around your protein. This is less precise, but also
>     less sensitive to local errors. If you want to know how your peptide
>     finds your protein, this is actually more informative anyway.
> 
>     There must be several programs that do this. I have done this for
>     MHC in the past with YASARA. It really explained nicely how he
>     peptide moved in.
> 
>      
> 
>     Cheers,
> 
>     Robbie
> 
>      
> 
>     Sent from my Windows 10 phone
> 
>      
> 
>     ------------------------------------------------------------------------
>     *From:* CCP4 bulletin board <[log in to unmask]
>     <mailto:[log in to unmask]>> on behalf of Dale Tronrud
>     <[log in to unmask] <mailto:[log in to unmask]>>
>     *Sent:* Friday, December 1, 2017 7:29:01 PM
>     *To:* [log in to unmask] <mailto:[log in to unmask]>
>     *Subject:* Re: [ccp4bb] Electrostatic Potential: Poisson-Boltzmann
>      
>        These are not easy questions to answer.  Certainly atoms,
>     particularly ones that are charged, even with fractional charges, have a
>     strong effect on the ESP.  If you delete them because you don't know
>     exactly where they are you will get a different answer than if you put
>     them in in some reasonable but unsupported location (as you have found).
>      This result indicates that the peptide does affect the ESP
>     significantly and you have to consider it.
> 
>        You could build lots of models with the peptide in different
>     conformations and average all the maps.  This misses the point.  You
>     have uncertainty in your model which means that you have uncertainty in
>     your electrostatic potential.  Any particular ESP that you calculate and
>     draw conclusions from will have a large uncertainty and you must
>     consider that uncertainty when deciding between your potential
>     conclusions.  (I'm not sure if the pun is intended or not!)
> 
>        I suppose you could believe that each possible conformation exists to
>     some extent in reality which means that all the ESP's you calculate
>     exist in some fraction of the molecules in the cell.  It is possible
>     that only the molecules with a particular conformation of this peptide
>     have the ESP that allows the molecule to function.  Life is hard.
> 
>        Another issue that you must consider: If the exact conformation of
>     this loop causes changes to the ESP that you consider significant to
>     your understanding of the function of this protein, the presence and
>     conformation of neighboring proteins and solvent will also cause
>     significant changes to the ESP.  The biological context of the protein
>     becomes important.  If your interpretation depends critically on the
>     value and distribution of ESP then I'm not sure you can work this out
>     based on calculated ESP, considering the large uncertainty.
> 
>     Dale Tronrud
> 
> 
>     On 12/1/2017 10:02 AM, chemocev marker wrote:
>     > Hi
>     > 
>     > I am calculating the Electrostatic Potential of my protein. But there
>     > were few flexible region with high B-factor and I deleted that part of
>     > the protein and then recalculated it. But there I can see a big
>     > change.As I have a structure in the presence and the absence of the
>     > peptide and the these flexible regions have a better map in the
>     > structure without peptide and with peptide I have to delete them.
>     > I have a question, should I model these missing regions
>     > 
>     > or
>     > 
>     > I should ignor them
>     > 
>     > best
>     > 
>     > Jiri
> 
>