Bug fixes

Note 1: Bug fixes are listed in reverse chronoligical order not in order of their importance
Note 2: Not all versions are saved. If a version was not distributed then it is not stored

Version Bug id Description Reported by

refmac 5.8 series

5.8.0085 ACE_C-N problem There was no planaraity and omega restraints on ACE - peptide links. Dictionary and the program have been updated to account for them. P.Kiser
5.8.0083 TRANS link problem There were problems with link directions. If TRANS link was defined between residues i1 and i2 with i2 < i1 then refmac would cobfuse links R.Joosten
Occupancy refinement: NO and NT During occupancy refinement if occupancies would become 0 then refmac would get confused. It hs now been fixed R.Joosten and J.Holton
5.8.0072 OXT problem refmac would replace OXT in RNA/DNA with OP3 which causes problem when modified RNA/DNA does have this atom R.Joosten
Quotes in merging When merging to dictionaries lincheck would forget quotes. R.Joosten
5.8.0071 TIP problem refmac would replace monomer name TIP with HOH. It causes problem when TIP is real ligand, e.g. as in 3eqn R.Joosten
5.8.0069 VDWR exclude There was a segmenttion fault when vdw restraint exclusion zones were defined by user. This bug was introduced while stacking restraints were implemented I.Tickle
5.8.0061 TWiN There was a problem in twin operator identification. Andrey Lebedev has now fixed it. R.Joosten
Jelly Body There was a problem with jelly body region selection. It has now been fixed by Keitaro Yamashita Keitaro Yamshita
5.8.0048 LINKS There was a problem with links on alternative positions. CCP4/APS workshop
LINKS there was a problem with symmetry related links introduced recently CCP4/APS workshop
5.8.0043 LINKS There was a problem with links introduced in version 5.8.0041. If hydrogens were added then remfac would ignore links. J.Tyndall
5.8.0041 Average Fo There was a problem in calculatin average Fo, Fc values. The number of reflections were miscalculated. E.Dodson, R.Nicholls
Rigid body There was a problem with rigid body. The problem was introduced while activating mmcif coordinate format reader. R.Joosten
5.8.0036 Symmetry related restraints There were some problems with symmetry related external restraints. They were not interpreted correctly. For example if a keyword was defined
external dist first .... symm yes
then program would not use symmetry. If symmetry was defined for each atom then it was used.
5.8.0035 No angle restraints The program would fail when there was no angle restraints. It is relevant for SAD only refinement C.Ballard
5.8.0033 Unrestrained atoms There was an infinity loop in some cases. It has now been fixed (this option is still not recommended unless you know what you are doing) Hideo Ago
5.8.0025 Report There was non-ascii character in the output related with reports on planaraity restraints R.Joosten
NMR target Minor bug fixes related with NMR target M.Rinaldelli
5.8.0024 Auto NCS Strictly speaking these are not bugs. They are more lik efeature enhancments. During alignment there were two hard coded parameters: 1) Maxumum chain length for alignment, it was 10 and 2) number of consecutive residues used for local RMS calculations, it was 5. Now these parameters can be controlled using following keywords:

ncsr align mlength [value]

# minumum length of chains that can be aligned. If the length of chain is smaller than this value then no alignmen ti sperformed

ncsr align rmslength [value]

Number of residues used for local alignment.

ncsr align mlength 10
ncsr align rmslength 5

refmac 5.7 series

5.7.0031 Links There was a problem with interpretation and use of pdb stype links (with oonly bond length). Now it is fixed and refmac uses distance restraints correspondign to pdb defined links B. Shaanon
Harmonic restraints report When harmonic restraints were used then refmac would not report info about their use. Now it does. Now refmac also can handle limited wild cards (e.g. if you define C* then all atoms with name starting with C will be understood) B.Shaanon
External restraints When MAKE HYDR YES was defined refmac would behave erratically with external restraints. E.g. it would not recognise perfectly valid atoms R.Nicholls, Yuval Mazor
B value restraints in the output pdb file Refmax 5.7 and later version would not write info about B value restraints to the output pdb file. Now it does. It should be remembered that post 5.7 versions of refmac use Kullback-Liebler divergence based B value restraints (effectively it means (B1-B2)^2/(B1*B2) with correspondign sigma). Outpud pdb file contains info about rms B value with equivalent sigma. J.Westbrook, D.Klenchin and many others
5.7.0030 sp2/sp3 hybradisation of N There were problems with dictionary generation especially for N containing ligands. N with three bonds were assumed to be sp2. It has now been corrected. If N has three bonds and it is bonded to sp2 atom then it is considered sp2, otherwise its hybradisation state is sp3. This affects libcheck and JLigand more than refmac -
5.7.0029 Transormation matrix There was a problem with conversion matrices for anisotropic B values. Fortunately it does not seem to affect refinement. It was a long standing problem. R.Nicholls
Map sharpeining It is more like new feature than a bug. Map sharpening would produce noisy map in some cases. Now sharpening has two components: 1) Removable B value and 2) regularised sharpening B value. Final sharpening map has coefficients: exp(B1 s^2/4) exp(-B2 s^2)/(exp(-2B2 s^2/4)+f(alpha,s)). B1 = Bmedian - (B_3rd_quartile-B_1st_quartile), B2 = min(Bmedian/2,B_3rd_quartile-B_1st_quartile) -
5.7.0028 Scaling problem In some cases refmac scaling would not converge and it would result in incorrect ML scale estimation, and it in turn would affect fom, maps etc causing refinement to halt or diverge. R.Joosten
5.7.0027 Mixed B value problem Now we have proper fix for the bug related with mixed B value refinement Eugene Orlov
DUM atoms Atom names for DUM atoms used by arp_warp were shifted to the right and it broke arp/warp. V.Lamzin
DUM atoms Record name for DUM atoms was HETATM. As V.Lamzin points out DUMs are not yet build protein atoms, so they should have ATOM record. V.Lamzin
Solvent mask output Refmac would stop if MSKOUT (write out mask) was defined. It was intentional. However as it was pointed out by K.Yamashita it would be desirable to continue refinement. Now refinemet continues and mask is written out at every cycle. If one wants to stop refinement then he/she can use ncycle 0 keyword. K.Yamashita
5.7.0026 Mixed B value refinement There is a bug when mixed anisotropic and isotropic refinement is performed. It appears on linux machines. Current fix is a temporary fix < Eugene Orlov
5.7.0023 Nested keywords Refmac would ignore if keywords files were nested: i.e. keywords file has @file1 and file 1 contains @file 2 etc. If file 2 contained comments (starting with #) then refmac would get confused and ignore all instructions in file 2 and further. R.Nicholls
5.7.0022 HETATM It is one of the long standing "bugs". REFMAC would not write out HETATM (all atoms had ATOM record) records. Now for ATOM record is used for standard amino acids and DNA/RNA. Many users
5.7.0021 FreeR segmentation Fault If in the input mtz file there were no reflection with flag 0 and default freeR flag was used then program would termintate with segmentation fault. Now, in these cases the program switches use of free R off -
5.7.0019 Symmetry links There was a problem in reading symmetry related links. This bug was introduced in 5.7.0018. The reason for this bug was change of the length torsion, planarity labels from 8 to 16. R.Joosten
SegFault There was a segmentation fault problem when hydrogens were added. Ths bug was introduced in 5.7.0018 version when occupancies of sp3 hydorgens on CB of ALA etc were asigned to one (before they were zer). R.Joosten
Geman-McClure parameters Whenever GM parameter was set to 0 refmac would change them to the default value. Now, if GM is set to less than 0 it is set to 0 R.Nicholls
5.7.0018 Hydrogens on sp3 atoms It could be considered as a bug or as a feature. Previous version of refmac did not (more precisely refmac would assign for occupancies of such hydrogens 0) put hydrogens on sp3 and sp2 atom when the number of missing hydorgens were 3 or 2 respectively. Now they are put in there "best" positions Many users
5.7.0017 Stop signal Stop signals did not work properly. Now it should work. Moreover you can use the followingcommands
stop F # do not stop unconditionally. It is redundant but may make clearer. Instead of F you can use 0 or N
stop T # stop undonditionally. Again it is redundant but may make things clearer. Instead of T one can use 1 or Y
Torsion labels Long torsion and plane labels (longer than 8 character) would cause the program to stop. It affected refmac and libcheck. Now the maximum length is for these labels is 16 T. Womack, R.Joosten, Yong Y Wang
5.7.0016 libcheck sigmas There were problems when libcheck combined two cif files.If sigma values where large (e.g. 100000) then the program would write series of * instead of actual value. T. Womack
5.7.0012 Symm link There were persisting problems with symmetry related links. Now it should work R.Joosten
5.7.0010 PDB link There were a problem in the interpretation and use of pdb defined links. It has now been fixed. If a user (pdb) has defined a link with a specified bond length then it will be used and the output pdb file will resemble the input pdb whre these links are present. -
libcheck In some version of the libcheck/JLigand would distort geometry for some ligands (it would happen to aromatic rings). Now it has been fixed. -
Free flag There were some problems with Free R flags when switching from phenix to refmac. Phenix uses by default freeR flag 1 and refmac uses 0. We have added some elementary analysis to recognise these cases and switch free R flag if possible. -
5.7.0009 Number of reflections Some of the previous version did not write the correct number of reflections to the pdb -
make_u_positive Now when B values (isotropic or anisotropic) trie to become negative refmacwrites info about them to the log file Many users
Problem with special positions has been fixed. If an atom close to itself by symmetry and it is not part of DNA/RNA peptide or sugar then it will be considered as on a special position and kept there Many users