New features for versions 1.10-2.10 of SOLVE
-
SOLVE now has easy scripts for SAD
phasing.
-
SOLVE has even better MAD phasing! SOLVE now re-refines
scattering factors using the final heavy atom parameters, improving the
final phases. In combination with version
1.04 or later of RESOLVE, the final maps are greatly improved over
version 1.17.
-
Must faster search for solutions:
SOLVE now follows only the very best solution by default, greatly speeding
up the search in most cases.
-
You can start from MR
or other input phases.
-
You can read CCP4 unmerged intensities
directly.
-
You can specify a SOLVETMPDIR where SOLVE
will write scratch files.
-
You can tell SOLVE where some sites are and go on from there easily with
ANALYZE_SOLVE
and with ADDSOLVE
-
There is now a SOLVE FAQS page.
-
SOLVE versions starting with 1.16 fix a serious bug that was present in
versions 1.10 to 1.15. This bug was that upon conversion from SOLVE's asymmetric
unit to the CCP4 asymmetric unit (done in routine "SOLVE" for all data,
whether mtz input or not) the phase of FA relative to Fp was not swapped
when the asymmetric unit changed. This affects the difference Fouriers
calculated by SOLVE for space groups where the CCP4 and SOLVE asymmetric
units for reflections are not the same.
-
SOLVE version 1.17 fixes some array dimensioning errors in versions up
to 1.16 that can cause it to hang when finding very large numbers (50 or
so) sites.
-
Solve version 1.17 sets the default of "resolution_steps" to 1 (it was
3 in previous versions). This improves MIR results for polar space groups
but could slow down Solve in some cases.
-
Solve version 1.18 had a minor bug in the way it dealt with SAD phasing,
always substituting in the default f-doubleprime values for the heavy atoms
from the tables, and including non-zero atomic scattering factors. Version
1.19 allows the user to input f-doubleprime (with fprprv_mad) and sets
the form factors other than f-doubleprime to zero. This creates the appropriate
scattering factors for a pseudo-derivative that is just like a native but
has anomalous differences. This pseudo-derivative is used for phasing with
the observed anomalous differences.
-
Solve version 1.18 had a dimension error preventing the use of more than
5 derivatives; this is fixed in 1.19.
-
Solve version 2.00 fixes a bug in previous versions of combine_all_data
where the number of wavelengths was not read in correctly for the second
dataset.
-
Solve version 2.00 has scattering factors for Br
-
Solve version 2.00 fixes bug which prevented solving SAD datasets in space
group P1.
-
Solve version 2.00 restores compatibility with new d*trek format.
-
Solve version 2.01 incorporates Sim weighting on the heavy-atom sites in
SAD phasing (this has the effect of replacing the heavy-atom sites in the
map, and for cases with a high proportion of anomalous scatterers such
as phosphorous phasing of DNA, makes a much better map).
-
Solve version 2.03 writes out fractional coordinates to solve.xyz and also
the inverse to solve_inverse.xyz.
-
Solve version 2.03 fixes a bug in SAD which resulted in no phases written
out if refinement of scattering factors failed.
-
RESOLVE version 2.03 fixes a bug in the output which caused the HL coefficients
from the input file to be copied without changes to the output file whenever
RESOLVE terminated without carrying out all requested cycles.
-
SOLVE version 2.03 changes the default BMIN (minimum B-value) from 15 to
2.0
-
RESOLVE version 2.03 includes iterative model-building and model rebuilding
-
RESOLVE version 2.04 will automatically figure out the optimal solvent
content and histograms for your structure
-
RESOLVE version 2.05 will use local patterns of density to improve the
map
-
RESOLVE version 2.05 has increased the weighting on map-based information
by a factor of 2 relative to earlier versions to match recalculated estimates
of the number of degrees of freedom in the map (now 2 * n_refl; was previously
n_refl).
-
RESOLVE version 2.06 introduces a new resolve_autobuild script that will
carry out pattern identification, fragment identification, and model-building,
using the superquick_build option, to get a partially refined model from
MAD/SAD/MIR data in just a few cycles.
-
RESOLVE version 2.08 fixes a bug which caused RESOLVE to fail when
NCS was found in the ha sites, but then later rejected due to lack of NCS
in the map.
-
The RESOLVE_BUILD script for version 2.08 uses the average of the top 5
structures built in the initial stages to start the rebuilding process.
-
RESOLVE version 2.08 introduces scoring of electron density maps based
on skew, correlation of local rms (as in SOLVE),
correlation of map calculated
with map-probability with original map, tertiary structure.
-
RESOLVE version 2.09 introduces greatly improved automated ligand fitting
-
RESOLVE version 2.09 fixes a problem with finding sites with MAD data in
trigonal/hexagonal space groups with 2-wavelength data or 4-wavelength data.
These searches will have generally failed with previous versions,
while single-wavelength or 3-wavelength data from the same datasets
would have succeeded. The phasing was fine for cases where the solution
was found, but the bug would normally prevent finding the heavy-atom sites.
There may be some datasets where version 2.09 can solve
2-wavelength data that could not be solved by previous versions.
(This problem was due to accidentally swapping the sign of anomalous
differences used in the MADMRG routine during the conversion to the
ccp4 asymmetric unit a number of times equal to the number of
wavelengths...thereby cancelling out the necessary swapping for
even-numbers of wavelengths...
The problem did not affect phasing, unless the "nobayes" flag was
specified, because phasing is done with all wavelengths, not with
the MADMRG data.
-
RESOLVE version 2.09 fixes a problem in reading heavy-atom sites and
problems with reading non-protein atoms in PDB files.
RESOLVE earlier versions would ignore the atom HG as a heavy-atom in
the NCS calculation, because the routines assumed that it was "H" (which
is ignored)ignored)
This is fixed in version 2.09, which now allows the use of
the element symbol specification in the PDB format to define the
atom type uniquely. (If no atom type is specified, in
SOLVE/RESOLVE the first character of the atom name is now used to
define the atom type for ATOM records, and all characters of the
atom name are used in HETATM records. This is all because otherwise it
is difficult to tell the difference between C-alpha and calcium atoms
without the element symbol specification, and many PDB files do
not contain element symbol specifications).
-
RESOLVE version 2.10 fixes a bug in which
a list can be overrun in model-building when the version is just big
enough to run. The symptom was the cryptic error message
"Please increase size of n_ca_max in iter_frag_place".
-
RESOLVE version 2.10 has a very powerful loop fitting algorithm that
you can run using a script or by using the new
PHENIX software.