Yes, you can provide your own bulk solvent model as a "partial 
structure" to REFMAC.
http://www.ysbl.york.ac.uk/~garib/refmac/docs/keywords/xray-principal.html#labin_fparti_phiparti
http://www.ysbl.york.ac.uk/~garib/refmac/docs/keywords/xray-general.html#scpa
http://www.ysbl.york.ac.uk/~garib/refmac/docs/keywords/xray-principal.html#solv

Once you have an electron density map for the bulk solvent that you are 
happy with, use the CCP4 program SFALL to convert that map into 
structure factors:

sfall mapin my_better_solvent.map hklout my_better_solvent.mtz << EOF
mode sfcalc mapin
EOF

  SFALL will output FC and PHIC, but since these names are sort of 
"taken" inside of REFMAC, you may want to re-name them to be Fsolv and 
PHIsolv using CAD.  While you are at it, you might as well combine Fsolv 
PHIsolv with your input MTZ file (the one that contains your observed 
data.  REFMAC only takes one input MTZ file.  For example:

cad hklin1 old_refme.mtz hklin2 my_better_solvent.mtz hklout 
new_refme.mtz << EOF
labin file 1 E1=FP E2=SIGFP E3=FreeR_flag
labin file 2 E1=FC E2=PHIC
labou file 2 E1=Fsolv E2=PHIsolv
scale file 2 1 0
EOF

Next, you want to alert REFMAC to the "partial structure" you are 
providing.  Do this by assigning them to FPART1 and PHIP1.  My input 
usually looks like this:

refmac5 hklin new_Refme.mtz xyzin best_model_yet.pdb hklout 
refmacout.mtz xyzout refmacout.pdb <<  EOF
LABIN FP=FP SIGFP=SIGFP FREE=FreeR_flag FPART1=Fsolv PHIP1=PHIsolv
SCPART 1
SOLVENT NO
EOF

The "SCPART 1" is required if you want REFMAC to refine a scale and B 
factor for your bulk solvent like it does with the bulk solvent it makes 
automatically.  If the map you input into SFALL used "1" to define a 
"solvent" map voxel and "0" to define vacuum, then the "scale factor" 
you will find under "Partial structure 1: scale" in the refmac output 
log will be your refined bulk solvent electron density (k_sol), which 
has units of electrons/Angstrom^3.

Note that I am using "SOLVENT NO" in the above example to turn off the 
"built-in" bulk solvent model.  If you don't do this, REFMAC will make 
its own bulk solvent as usual, but refine it as an additional partial 
structure.  Then you will have two bulk solvent masks!   REFMAC actually 
supports multiple partial structures (up to four, I think), but my 
experience with this is that it gets unstable rather quickly.

That's the mechanics of it anyway.  It is a pity that phenix.refine 
doesn't support external partial structures like this, but perhaps it 
will soon.

One thing to beware is that MSKOUT may or may not be on any particular 
scale.  If you want to actually recover the "refined" bulk solvent map, 
you need to first re-scale the mskout.map so that it ranges from 0 to 1 
in value (check this with MAPDUMP), before you feed it to SFALL.  Then, 
during the CAD step above apply an appropriate scale and B factor to 
Fsolv using "SCALE FILE 1":
http://www.ccp4.ac.uk/html/cad.html#scale
You will find appropriate scale and B factor values printed out on the 
last "Partial structure   1:" entry in your last refmac log.


Now, of course, you have the finer detail problem of what, exactly, to 
do to the "default" bulk solvent mask to make it better.   Simply adding 
the Fo-Fc map is not only cheating; it doesn't work!  By default, the 
Fo-Fc map does not contain structure factors flagged as part of the FREE 
set.  So, if you "edit" your bulk solvent by adding Fo-Fc, you will 
immediately get an Rwork of zero, but will see no change to Rfree.  This 
is because all the "free" structure factors used to make the Fo-Fc map 
were zero.

Finding some way to "transfer" information from the differences in the 
working set into the free set is challenging, and indeed this is the 
central problem of model building.

-James Holton
MAD Scientist

On 1/11/2015 11:05 AM, Alastair Fyfe wrote:
> A related/follow-on question, hopefully on the same topic:
>
> When mask-based solvent modeling leads to problems caused by 
> calculation of an inappropriate mask (eg inclusion of disordered loops 
> or inaccessible pockets as bulk-solvent volume) it seems that a 
> feasible workaround is to (a) output the calculated mask via MSKOUT 
> (b) edit it to remove spurious regions (c) calculate partial structure 
> factors for the new solvent region using refmac's previous kSol, bSol 
> estimates (d)  include the partial structure factors in a second round 
> of refinement by specifying "SOLVENT NO", FPART and SCPART.
>
> However, this does not re-calculate kSol and bSol to reflect the 
> modified mask. Is there a way to simply supply the mask to be used 
> rather than have refmac calculate it ?
> Thanks,
> Alastair Fyfe
>
>
> On 01/11/2015 08:04 AM, Eleanor Dodson wrote:
>> Yes. If the model is incomplete it is obviously not sensible to use the
>> mask based solvent - you will tend to lose the unmodelled features. 
>> It also
>> gives unrealistically low R factors for a crystal with high solvent
>> content. However the best test would be to analyse maps and try to 
>> decide
>> if & when the different procedures work best.. A project for a student
>> dissertation perhaps??
>>    Eleanor
>>
>> On 9 January 2015 at 20:13, Roberts, Sue A - (suer) 
>> <[log in to unmask]>
>> wrote:
>>
>>> I always try both methods - usually there is little difference.  
>>> However,
>>>
>>> For CueO (multicopper oxidase) where there are about  25 disordered
>>> (unseen) residues in a loop,  using Babinet scaling instead of the 
>>> default
>>> refmac scaling reduced the R factors by about 2% (both R and Rfree) and
>>> improved the quality of the maps substantially.
>>>
>>>  From Dirk's comment, I'd guess this is because the mask-based solvent
>>> model is putting solvent where there is (disordered) protein, which is
>>> different from real bulk solvent.
>>>
>>> Sue
>>>
>>> Dr. Sue A. Roberts
>>> Dept. of Chemistry and Biochemistry
>>> University of Arizona
>>> 1306 E. University Blvd,  Tucson, AZ 85721
>>> Phone: 520 621 4168
>>> [log in to unmask]
>>>
>>>
>>>
>>> -----Original Message-----
>>> From: CCP4 bulletin board [mailto:[log in to unmask]] On Behalf Of
>>> Armando Albert
>>> Sent: Friday, January 09, 2015 12:56 AM
>>> To: [log in to unmask]
>>> Subject: [ccp4bb] Bulk solvent
>>>
>>> Dear all,
>>> Is there any reason for using Babinet scaling for bulk solvent 
>>> correction
>>> instead of mask based scaling?
>>> Armando
>>>