[phenixbb] phenix.polder - tool for calculating omit maps by excluding bulk solvent

Pavel Afonine pafonine at lbl.gov
Thu Apr 21 08:57:20 PDT 2016

Hi Tim,

thanks for bringing this! Frankly I was not aware of SQUEEZE method 
likely because I'm not a heavy user of Platon. Anyways, this 
conversation prompted me to do some reading this morning:


Here is what I gather from this. Well, first we are talking about 
small-molecule crystallography here, with all the implications. It may 
happen that crystals contain areas occupied by disordered solvent that 
is impractical to account for in terms of atomic model. I'm not an 
expert in small molecule crystallography but it sounds to me like this 
is something that rather does not happen all that often compared to 
bio-crystallography where the disordered (bulk) solvent typically 
occupies from 10 to 90% of the unit cell volume. Now, given high 
accuracy standards in small molecule crystallography this inability to 
account for disordered scattering poses a big problem as it hampers 
refinement and potentially highly accurate map interpretation. The 
SQUEEZE method provides a way to identify such regions of disordered 
scatterers and accounting for their contribution to the total model 
structure factors. For example, citing

A. L. Spek
Acta Cryst. (2015). C71, 9-18
  PLATON SQUEEZE: a tool for the calculation of the disordered solvent 
contribution to the calculated structure factors

The completion of a crystal structure determination is often hampered by 
the presence of embedded solvent molecules or ions that are seriously 
disordered. Their contribution to the calculated structure factors in 
the least-squares refinement of a crystal structure has to be included 
in some way. Traditionally, an atomistic solvent disorder model is 
attempted. Such an approach is generally to be preferred, but it does 
not always lead to a satisfactory result and may even be impossible in 
cases where channels in the structure are filled with continuous 
electron density. This paper documents the SQUEEZE method as an 
alternative means of addressing the solvent disorder issue. It 
conveniently interfaces with the 2014 version of the least-squares 
refinement program SHELXL [Sheldrick (2015). Acta Cryst. C71. In the 
press] and other refinement programs that accept externally provided 
fixed contributions to the calculated structure factors. The PLATON 
SQUEEZE tool calculates the solvent contribution to the structure 
factors by back-Fourier transformation of the electron density found in 
the solvent-accessible region of a phase-optimized difference 
electron-density map. The actual least-squares structure refinement is 
delegated to, for example, SHELXL.

Here is what we are dealing with in our case. Macro-molecular crystals 
on average contain ~50% of the disordered (bulk) solvent. Most (if not 
all) software packages automatically account for this disordered solvent 
by defining the total model structure factors as

Fmodel = k_total * (Fcalc_atoms + Fbulk) .

Fmodel is then used in all calculations such as R-factors, refinement 
targets, various maps, etc. Up to this point, it is along the lines of 
what SQUEEZE does, indeed.

Different bulk-solvent models can be used to calculate Fbulk 
contribution. Two major models are in use: Babinet-based model (used in 
SHELX) and Flat model (used in CNS, REFMAC, PHENIX).

Both models have their pros and cons. For example, the downside of 
Babinet-based model is that it holds true for resolutions lower than 10-15A:

Podjarny, A. D. & Urzhumtsev, A. G. (1997).

and is handicapped at resolutions between 10-15 and 5-6A (where 
disordered solvent contribution vanishes). The good thing about it is 
that it does not implies masking bias.

Unlike Babinet model, Flat bulk-solvent model accounts for disordered 
solvent pretty well across all resolution ranges. The downside of the 
flat bulk solvent model is what we are trying to address using Polder maps.

So.. The way flat bulk solvent model works is it defines a solvent mask 
which is a binary function with 0 inside macro-molecule and 1 outside 
(Jiang&Brunger, 1997). Then this function is Fourier transformed into 
structure factors Fmask and that are then added to the total model 
structure factors with some refinable scale k_mask:

Fmodel = k_total * (Fcalc_atoms + k_mask * Fmask) .

The problem with this approach is that the solvent is "poured" 
everywhere in the unit cell where there is no atomic model. For example, 
if there is a ligand that is not placed yet or a loop that is not 
modeled yet, the flat bulk solvent will fill the gap. Most of the time 
this will not pose much trouble as atomic features typically stand above 
the noise or/and solvent density. However, in cases when feature that 
one tries to model is weak (mobile, partially occupied ligand or 
flexible disordered loop) the flat solvent model may obscure it by 
flattening corresponding density in the region of interest. This is the 
issue that Polder residual OMIT map is meant to address by excluding 
bulk-solvent contribution from specifically defined regions and 
therefore provide mask bias free view of residual map in that region. I 
think this is not quite the same as what SQUEEZE method does.

In fact Polded OMIT map is a single iteration of a more general 
procedure described in Section 2.4 and Figure 6 here:


All the best,

On 4/21/16 06:21, Tim Gruene wrote:
> Dear Dorothee,
> With squeeze, you remove solvent in order to make features visible lying
> underneath the noise density of the solvent. That's reminiscent to me as the
> cartoon on p.9 of the phenix_polder.pdf
> In Platon, the structure factors are calculated from the density in the
> solvent region, Eq. (4) in the Platon paper. That appears to be the same as
> explained for phenix_polder on p. 4, except that phenix replaces
> rho(x_solvent) with 1 for the mask. The equation on p.2 of the PDF-file is
> identical to the line below Eq. 4 in the SQUEEZE paper, so it seems
> conceptually pretty much the same to me.
> Since SQUEEZE was presented at the ACA in 1998 (based on a paper from 1990), I
> thought you may have been motivated by it. It is probably not much surprising
> that good ideas get invented at various places.
> Best wishes,
> Tim
> On Wednesday, April 20, 2016 01:59:34 PM you wrote:
>> Hi Tim,
>> I quickly looked over the SQUEEZE command in PLATON (are you referring to:
>> http://scripts.iucr.org/cgi-bin/paper?S2053229614024929 ?). To me, it does
>> not seem to be related to phenix.polder.
>> There is no complicated math involved in polder; slides 2-4 are a summary
>> of the flat bulk-solvent model (which is used in Phenix, and which is also
>> available in CNS and REFMAC).
>> The flat bulk-solvent model is described first here:
>> Phillips, S. E. (1980). *J. Mol. Biol.* *142*, 531–554.
>> I uses a similar concept than SQUEEZE, i.e. the total structure factor is
>> expressed as a sum of contributions from protein model and disordered
>> solvent.
>> More references can be found in this review about bulk solvent models in MX:
>> Weichenberger, C. X., Afonine, P. V, Kantardjieff, K. & Rupp, B. (2015).
>> *Acta Crystallogr. Sect. D Biol. Crystallogr.* *71*, 1023–1038.
>> The polder tool uses the bulk solvent mask (as it is generated for other
>> functionalities in phenix, such as phenix.refine), and then modifies the
>> mask locally. I am sorry if the presentation file is misleading, I should
>> maybe add some references to make clear what is summary and what is related
>> to the polder tool.
>> Best wishes,
>> Dorothee
>> PS:
>> I did not understand how the name "squeeze" relates to "polder"...
>> On Wed, Apr 20, 2016 at 12:07 PM, Tim Gruene <tim.gruene at psi.ch> wrote:
>>> Hash: SHA1
>>> Hi Pavel,
>>> this is an interesting concept. It seems related to the SQUEEZE command in
>>> platon - even the name appears to suggest a relationship. I did not
>>> understand
>>> the maths entirely: Are they similar concept, or do I misinterprete?
>>> Best,
>>> Tim
>>> On Wednesday, April 20, 2016 07:28:16 AM Pavel Afonine wrote:
>>>> Hello,
>>>> it's good to know it was useful for you, thanks for feedback! I afraid
>>>> it's too new so that we don't have a publication to cite yet. We are
>>>> working on a manuscript but it may take a little while before it appears
>>>> somewhere. For now I guess you can use this link (unless Dorothee has a
>>>> better idea)
>>>> http://www.phenix-online.org/presentations/phenix_polder.pdf
>>>> and use official Phenix citation:
>>>> Acta Cryst. D66, 213-221 (2010).
>>>> That's all we have at the moment anyway.
>>>> All the best,
>>>> Pavel
>>>> On 4/20/16 04:18, Lund Bjarte Aarmo wrote:
>>>>> DearDorothee and phenixbb,
>>>>> I found this software very useful for protein-fragment complexes with
>>>>> weak electron density. I was wondering how the software should be
>>> cited?
>>>>> Kind regards,
>>>>> Bjarte Aarmo Lund
>>>>> PhD candidate
>>>>> UiT – The arctic university of Norway
>>>>> *From:*phenixbb-bounces at phenix-online.org
>>>>> [mailto:phenixbb-bounces at phenix-online.org] *On Behalf Of *Dorothee
>>>>> Liebschner
>>>>> *Sent:* 22. mars 2016 21:46
>>>>> *To:* PHENIX user mailing list <phenixbb at phenix-online.org>
>>>>> *Subject:* [phenixbb] phenix.polder - tool for calculating omit maps
>>>>> by excluding bulk solvent
>>>>> Dear phenix users,
>>>>> Starting from the nightly build dev-2356, a new tool for calculating
>>>>> ligand omit-maps, called 'polder', is included in phenix.
>>>>> Usage:
>>>>> phenix.polder model.pdb    data.mtz    selection='chain A and resseq
>>> 123’
>>>>> Phenix.polder calculates omit maps for atom selections by preventing
>>>>> the bulk solvent mask to flood into the atom selection area and its
>>>>> vicinity. The tool can be useful in cases where the density of the
>>>>> selected atoms is weak and possibly obscured by bulk solvent.
>>>>> Polder produces less biased maps compared to procedures where the atom
>>>>> selection occupancy is set to zero, and the atoms are included in the
>>>>> solvent mask calculation (in that case, the resulting difference
>>>>> density can have similar shape than the selected atoms). Phenix.polder
>>>>> excludes a larger volume from the bulk solvent and therefore prevents
>>>>> misinterpreting bulk solvent density as omit density.
>>>>> If you want to know more about how the tool is working and to see some
>>>>> examples, have a look at the presentation file:
>>>>> https://www.phenix-online.org/presentations/phenix_polder.pdf.
>>>>> The documentation page can be found here:
>>>>> www.phenix-online.org/version_docs/dev-2356/reference/polder.html
>>>>> Best wishes,
>>>>> Dorothee
>>>>> _______________________________________________
>>>>> phenixbb mailing list
>>>>> phenixbb at phenix-online.org
>>>>> http://phenix-online.org/mailman/listinfo/phenixbb
>>>>> Unsubscribe: phenixbb-leave at phenix-online.org
>>> - --
>>> - --
>>> Paul Scherrer Institut
>>> Tim Gruene
>>> - - persoenlich -
>>> OFLC/102
>>> CH-5232 Villigen PSI
>>> phone: +41 (0)56 310 5297
>>> GPG Key ID = A46BEE1A
>>> Version: GnuPG v2
>>> iD8DBQFXF9NpUxlJ7aRr7hoRApIFAKCZ6gKgFMeeU4DSZfr5i/5Z3inC/ACeOmIh
>>> sjQCpCYG2CkH/N/RoIfhApQ=
>>> =t+dS
>>> -----END PGP SIGNATURE-----
>>> _______________________________________________
>>> phenixbb mailing list
>>> phenixbb at phenix-online.org
>>> http://phenix-online.org/mailman/listinfo/phenixbb
>>> Unsubscribe: phenixbb-leave at phenix-online.org
>>> _______________________________________________
>>> phenixbb mailing list
>>> phenixbb at phenix-online.org
>>> http://phenix-online.org/mailman/listinfo/phenixbb
>>> Unsubscribe: phenixbb-leave at phenix-online.org

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://phenix-online.org/pipermail/phenixbb/attachments/20160421/86d6b853/attachment-0001.htm>

More information about the phenixbb mailing list