WHAT IF Check report

This file was created 2011-12-15 from WHAT_CHECK output by a conversion script. If you are new to WHAT_CHECK, please study the pdbreport pages. There also exists a legend to the output.

Please note that you are looking at an abridged version of the output (all checks that gave normal results have been removed from this report). You can have a look at the Full report instead.

Verification log for pdb3f9b.ent

Checks that need to be done early-on in validation

Warning: Topology could not be determined for some ligands

Some ligands in the table below are too complicated for the automatic topology determination. WHAT IF uses a local copy of Daan van Aalten's Dundee PRODRG server to automatically generate topology information for ligands. Some molecules are too complicated for this software. If that happens, WHAT IF / WHAT-CHECK continue with a simplified topology that lacks certain information. Ligands with a simplified topology can, for example, not form hydrogen bonds, and that reduces the accuracy of all hydrogen bond related checking facilities.

The reason for topology generation failure is indicated. 'Atom types' indicates that the ligand contains atom types not known to PRODRUG. 'Attached' means that the ligand is covalently attached to a macromolecule. 'Size' indicates that the ligand has either too many atoms (or two or less which PRODRUG also cannot cope with), or too many bonds, angles, or torsion angles. 'Fragmented' is written when the ligand is not one fully covalently connected molecule but consists of multiple fragments. 'N/O only' is given when the ligand contains only N and/or O atoms. 'OK' indicates that the automatic topology generation succeeded.

 284 VO4   ( 501-)  A  -         Atom types
 285 PDV   ( 502-)  A  -         Atom types
 286 GOL   ( 503-)  A  -         OK
 287 GOL   ( 504-)  A  -         OK

Non-validating, descriptive output paragraph

Note: Ramachandran plot

In this Ramachandran plot x-signs represent glycines, squares represent prolines, and plus-signs represent the other residues. If too many plus- signs fall outside the contoured areas then the molecule is poorly refined (or worse). Proline can only occur in the narrow region around phi=-60 that also falls within the other contour islands.

In a colour picture, the residues that are part of a helix are shown in blue, strand residues in red. Preferred regions for helical residues are drawn in blue, for strand residues in red, and for all other residues in green. A full explanation of the Ramachandran plot together with a series of examples can be found at the WHAT_CHECK website.

Chain identifier: A

Coordinate problems, unexpected atoms, B-factor and occupancy checks

Warning: What type of B-factor?

WHAT IF does not yet know well how to cope with B-factors in case TLS has been used. It simply assumes that the B-factor listed on the ATOM and HETATM cards are the total B-factors. When TLS refinement is used that assumption sometimes is not correct. The header of the PDB file states that TLS groups were used. So, if WHAT IF complains about your B-factors, while you think that they are OK, then check for TLS related B-factor problems first.

Obviously, the temperature at which the X-ray data was collected has some importance too:


Number of TLS groups mentione in PDB file header: 0

Crystal temperature (K) :100.000

Note: B-factor plot

The average atomic B-factor per residue is plotted as function of the residue number.

Chain identifier: A

Nomenclature related problems

Warning: Arginine nomenclature problem

The arginine residues listed in the table below have their N-H-1 and N-H-2 swapped.

  44 ARG   ( 230-)  A
  86 ARG   ( 272-)  A
 151 ARG   ( 337-)  A

Warning: Tyrosine convention problem

The tyrosine residues listed in the table below have their chi-2 not between -90.0 and 90.0

   3 TYR   ( 189-)  A
  62 TYR   ( 248-)  A
  75 TYR   ( 261-)  A
 122 TYR   ( 308-)  A
 197 TYR   ( 383-)  A

Warning: Phenylalanine convention problem

The phenylalanine residues listed in the table below have their chi-2 not between -90.0 and 90.0.

  43 PHE   ( 229-)  A
  85 PHE   ( 271-)  A
 110 PHE   ( 296-)  A

Warning: Aspartic acid convention problem

The aspartic acid residues listed in the table below have their chi-2 not between -90.0 and 90.0, or their proton on OD1 instead of OD2.

  45 ASP   ( 231-)  A
  57 ASP   ( 243-)  A
 114 ASP   ( 300-)  A
 144 ASP   ( 330-)  A
 207 ASP   ( 393-)  A
 266 ASP   ( 452-)  A

Warning: Glutamic acid convention problem

The glutamic acid residues listed in the table below have their chi-3 outside the -90.0 to 90.0 range, or their proton on OE1 instead of OE2.

  38 GLU   ( 224-)  A
  82 GLU   ( 268-)  A
  90 GLU   ( 276-)  A
 104 GLU   ( 290-)  A
 177 GLU   ( 363-)  A
 191 GLU   ( 377-)  A
 263 GLU   ( 449-)  A
 273 GLU   ( 459-)  A

Geometric checks

Warning: Unusual bond lengths

The bond lengths listed in the table below were found to deviate more than 4 sigma from standard bond lengths (both standard values and sigmas for amino acid residues have been taken from Engh and Huber [REF], for DNA they were taken from Parkinson et al [REF]). In the table below for each unusual bond the bond length and the number of standard deviations it differs from the normal value is given.

Atom names starting with "-" belong to the previous residue in the chain. If the second atom name is "-SG*", the disulphide bridge has a deviating length.

 172 THR   ( 358-)  A      CB   OG1   1.52    5.5

Warning: Low bond length variability

Bond lengths were found to deviate less than normal from the mean Engh and Huber [REF] and/or Parkinson et al [REF] standard bond lengths. The RMS Z-score given below is expected to be near 1.0 for a normally restrained data set. The fact that it is lower than 0.667 in this structure might indicate that too-strong restraints have been used in the refinement. This can only be a problem for high resolution X-ray structures.

RMS Z-score for bond lengths: 0.384
RMS-deviation in bond distances: 0.009

Warning: Low bond angle variability

Bond angles were found to deviate less than normal from the standard bond angles (normal values for protein residues were taken from Engh and Huber [REF], for DNA/RNA from Parkinson et al [REF]). The RMS Z-score given below is expected to be near 1.0 for a normally restrained data set. The fact that it is lower than 0.667 in this structure might indicate that too-strong restraints have been used in the refinement. This can only be a problem for high resolution X-ray structures.

RMS Z-score for bond angles: 0.575
RMS-deviation in bond angles: 1.198

Error: Nomenclature error(s)

Checking for a hand-check. WHAT IF has over the course of this session already corrected the handedness of atoms in several residues. These were administrative corrections. These residues are listed here.

  38 GLU   ( 224-)  A
  44 ARG   ( 230-)  A
  45 ASP   ( 231-)  A
  57 ASP   ( 243-)  A
  82 GLU   ( 268-)  A
  86 ARG   ( 272-)  A
  90 GLU   ( 276-)  A
 104 GLU   ( 290-)  A
 114 ASP   ( 300-)  A
 144 ASP   ( 330-)  A
 151 ARG   ( 337-)  A
 177 GLU   ( 363-)  A
 191 GLU   ( 377-)  A
 207 ASP   ( 393-)  A
 263 GLU   ( 449-)  A
 266 ASP   ( 452-)  A
 273 GLU   ( 459-)  A

Torsion-related checks

Warning: Torsion angle evaluation shows unusual residues

The residues listed in the table below contain bad or abnormal torsion angles.

These scores give an impression of how `normal' the torsion angles in protein residues are. All torsion angles except omega are used for calculating a `normality' score. Average values and standard deviations were obtained from the residues in the WHAT IF database. These are used to calculate Z-scores. A residue with a Z-score of below -2.0 is poor, and a score of less than -3.0 is worrying. For such residues more than one torsion angle is in a highly unlikely position.

 259 VAL   ( 445-)  A    -2.1

Warning: Backbone evaluation reveals unusual conformations

The residues listed in the table below have abnormal backbone torsion angles.

Residues with `forbidden' phi-psi combinations are listed, as well as residues with unusual omega angles (deviating by more than 3 sigma from the normal value). Please note that it is normal if about 5 percent of the residues is listed here as having unusual phi-psi combinations.

  35 CYS   ( 221-)  A  omega poor
  61 ASN   ( 247-)  A  omega poor
 114 ASP   ( 300-)  A  Poor phi/psi, omega poor
 132 THR   ( 318-)  A  Poor phi/psi
 139 ASP   ( 325-)  A  Poor phi/psi
 146 TYR   ( 332-)  A  omega poor
 169 PRO   ( 355-)  A  omega poor
 171 GLN   ( 357-)  A  omega poor
 211 LEU   ( 397-)  A  Poor phi/psi
 217 CYS   ( 403-)  A  Poor phi/psi
 240 GLN   ( 426-)  A  Poor phi/psi
 258 MET   ( 444-)  A  omega poor
 259 VAL   ( 445-)  A  Poor phi/psi
 chi-1/chi-2 correlation Z-score : 0.812

Warning: Unusual rotamers

The residues listed in the table below have a rotamer that is not seen very often in the database of solved protein structures. This option determines for every residue the position specific chi-1 rotamer distribution. Thereafter it verified whether the actual residue in the molecule has the most preferred rotamer or not. If the actual rotamer is the preferred one, the score is 1.0. If the actual rotamer is unique, the score is 0.0. If there are two preferred rotamers, with a population distribution of 3:2 and your rotamer sits in the lesser populated rotamer, the score will be 0.667. No value will be given if insufficient hits are found in the database.

It is not necessarily an error if a few residues have rotamer values below 0.3, but careful inspection of all residues with these low values could be worth it.

  13 SER   ( 199-)  A    0.37

Warning: Unusual backbone conformations

For the residues listed in the table below, the backbone formed by itself and two neighbouring residues on either side is in a conformation that is not seen very often in the database of solved protein structures. The number given in the table is the number of similar backbone conformations in the database with the same amino acid in the centre.

For this check, backbone conformations are compared with database structures using C-alpha superpositions with some restraints on the backbone oxygen positions.

A residue mentioned in the table can be part of a strange loop, or there might be something wrong with it or its directly surrounding residues. There are a few of these in every protein, but in any case it is worth looking at!

  26 THR   ( 212-)  A      0
  27 ASN   ( 213-)  A      0
  31 TYR   ( 217-)  A      0
  33 GLN   ( 219-)  A      0
  34 ALA   ( 220-)  A      0
  35 CYS   ( 221-)  A      0
  36 GLY   ( 222-)  A      0
  40 LEU   ( 226-)  A      0
  41 ASN   ( 227-)  A      0
  42 ARG   ( 228-)  A      0
  43 PHE   ( 229-)  A      0
  46 ILE   ( 232-)  A      0
  50 ARG   ( 236-)  A      0
  54 VAL   ( 240-)  A      0
  55 ARG   ( 241-)  A      0
  59 ASN   ( 245-)  A      0
  60 ALA   ( 246-)  A      0
  65 VAL   ( 251-)  A      0
  67 ASN   ( 253-)  A      0
  93 THR   ( 279-)  A      0
  94 PRO   ( 280-)  A      0
  99 LEU   ( 285-)  A      0
 108 GLN   ( 294-)  A      0
 115 TYR   ( 301-)  A      0
 122 TYR   ( 308-)  A      0
And so on for a total of 99 lines.

Warning: Backbone oxygen evaluation

The residues listed in the table below have an unusual backbone oxygen position.

For each of the residues in the structure, a search was performed to find 5-residue stretches in the WHAT IF database with superposable C-alpha coordinates, and some restraining on the neighbouring backbone oxygens.

In the following table the RMS distance between the backbone oxygen positions of these matching structures in the database and the position of the backbone oxygen atom in the current residue is given. If this number is larger than 1.5 a significant number of structures in the database show an alternative position for the backbone oxygen. If the number is larger than 2.0 most matching backbone fragments in the database have the peptide plane flipped. A manual check needs to be performed to assess whether the experimental data can support that alternative as well. The number in the last column is the number of database hits (maximum 80) used in the calculation. It is "normal" that some glycine residues show up in this list, but they are still worth checking!

  66 GLY   ( 252-)  A   1.53   12

Bump checks

Error: Abnormally short interatomic distances

The pairs of atoms listed in the table below have an unusually short interactomic distance; each bump is listed in only one direction.

The contact distances of all atom pairs have been checked. Two atoms are said to `bump' if they are closer than the sum of their Van der Waals radii minus 0.40 Angstrom. For hydrogen bonded pairs a tolerance of 0.55 Angstrom is used. The first number in the table tells you how much shorter that specific contact is than the acceptable limit. The second distance is the distance between the centres of the two atoms. Although we believe that two water atoms at 2.4 A distance are too close, we only report water pairs that are closer than this rather short distance.

The last text-item on each line represents the status of the atom pair. If the final column contains the text 'HB', the bump criterion was relaxed because there could be a hydrogen bond. Similarly relaxed criteria are used for 1-3 and 1-4 interactions (listed as 'B2' and 'B3', respectively). BL indicates that the B-factors of the clashing atoms have a low B-factor thereby making this clash even more worrisome. INTRA and INTER indicate whether the clashes are between atoms in the same asymmetric unit, or atoms in symmetry related asymmetric units, respectively.

 172 THR   ( 358-)  A      CG2 <->  173 ALA   ( 359-)  A      N      0.27    2.73  INTRA
 172 THR   ( 358-)  A      CG2 <->  288 HOH   ( 157 )  A      O      0.25    2.55  INTRA
 234 ASN   ( 420-)  A      ND2 <->  288 HOH   (  88 )  A      O      0.24    2.46  INTRA
 168 PHE   ( 354-)  A      CE1 <->  223 ARG   ( 409-)  A      CD     0.08    3.12  INTRA
  39 LYS   ( 225-)  A      NZ  <->  288 HOH   ( 610 )  A      O      0.08    2.62  INTRA BF
 223 ARG   ( 409-)  A      CG  <->  285 PDV   ( 502-)  A      O5     0.07    2.73  INTRA
  69 ARG   ( 255-)  A      NE  <->  208 ASP   ( 394-)  A      OD1    0.06    2.64  INTRA
 115 TYR   ( 301-)  A      OH  <->  216 HIS   ( 402-)  A      NE2    0.05    2.65  INTRA BL
 223 ARG   ( 409-)  A      N   <->  285 PDV   ( 502-)  A      O3     0.05    2.65  INTRA
  35 CYS   ( 221-)  A      N   <->   36 GLY   ( 222-)  A      CA     0.04    2.86  INTRA
   1 SER   ( 187-)  A      N   <->    2 PRO   ( 188-)  A      CD     0.03    2.97  INTRA
 108 GLN   ( 294-)  A      NE2 <->  288 HOH   ( 617 )  A      O      0.02    2.68  INTRA
  35 CYS   ( 221-)  A      SG  <->  288 HOH   ( 539 )  A      O      0.02    2.98  INTRA
 232 CYS   ( 418-)  A      SG  <->  246 MET   ( 432-)  A      SD     0.01    3.44  INTRA
 194 ARG   ( 380-)  A      NH2 <->  211 LEU   ( 397-)  A      O      0.01    2.69  INTRA
 114 ASP   ( 300-)  A      OD1 <->  288 HOH   ( 575 )  A      O      0.01    2.39  INTRA

Packing, accessibility and threading

Note: Inside/Outside RMS Z-score plot

The Inside/Outside distribution normality RMS Z-score over a 15 residue window is plotted as function of the residue number. High areas in the plot (above 1.5) indicate unusual inside/outside patterns.

Chain identifier: A

Warning: Abnormal packing environment for some residues

The residues listed in the table below have an unusual packing environment.

The packing environment of the residues is compared with the average packing environment for all residues of the same type in good PDB files. A low packing score can indicate one of several things: Poor packing, misthreading of the sequence through the density, crystal contacts, contacts with a co-factor, or the residue is part of the active site. It is not uncommon to see a few of these, but in any case this requires further inspection of the residue.

 171 GLN   ( 357-)  A      -7.01
   3 TYR   ( 189-)  A      -5.92
  35 CYS   ( 221-)  A      -5.83
 109 ARG   ( 295-)  A      -5.80
 275 GLN   ( 461-)  A      -5.50
 118 GLN   ( 304-)  A      -5.37
 234 ASN   ( 420-)  A      -5.10

Note: Quality value plot

The quality value smoothed over a 10 residue window is plotted as function of the residue number. Low areas in the plot (below -2.0) indicate unusual packing.

Chain identifier: A

Warning: Low packing Z-score for some residues

The residues listed in the table below have an unusual packing environment according to the 2nd generation packing check. The score listed in the table is a packing normality Z-score: positive means better than average, negative means worse than average. Only residues scoring less than -2.50 are listed here. These are the unusual residues in the structure, so it will be interesting to take a special look at them.

 205 VAL   ( 391-)  A   -2.64
  51 GLN   ( 237-)  A   -2.60

Warning: Abnormal packing Z-score for sequential residues

A stretch of at least four sequential residues with a 2nd generation packing Z-score below -1.75 was found. This could indicate that these residues are part of a strange loop or that the residues in this range are incomplete, but it might also be an indication of misthreading.

The table below lists the first and last residue in each stretch found, as well as the average residue Z-score of the series.

 203 SER   ( 389-)  A     -  206 ALA   ( 392-)  A        -1.98

Note: Second generation quality Z-score plot

The second generation quality Z-score smoothed over a 10 residue window is plotted as function of the residue number. Low areas in the plot (below -1.3) indicate unusual packing.

Chain identifier: A

Water, ion, and hydrogenbond related checks

Error: HIS, ASN, GLN side chain flips

Listed here are Histidine, Asparagine or Glutamine residues for which the orientation determined from hydrogen bonding analysis are different from the assignment given in the input. Either they could form energetically more favourable hydrogen bonds if the terminal group was rotated by 180 degrees, or there is no assignment in the input file (atom type 'A') but an assignment could be made. Be aware, though, that if the topology could not be determined for one or more ligands, then this option will make errors.

 164 HIS   ( 350-)  A

Warning: Buried unsatisfied hydrogen bond donors

The buried hydrogen bond donors listed in the table below have a hydrogen atom that is not involved in a hydrogen bond in the optimized hydrogen bond network.

Hydrogen bond donors that are buried inside the protein normally use all of their hydrogens to form hydrogen bonds within the protein. If there are any non hydrogen bonded buried hydrogen bond donors in the structure they will be listed here. In very good structures the number of listed atoms will tend to zero.

Waters are not listed by this option.

  95 VAL   ( 281-)  A      N
 100 ALA   ( 286-)  A      N
 118 GLN   ( 304-)  A      N
 188 GLN   ( 374-)  A      NE2
 202 SER   ( 388-)  A      OG
 218 ARG   ( 404-)  A      NE
 219 ALA   ( 405-)  A      N
 221 VAL   ( 407-)  A      N
 222 GLY   ( 408-)  A      N
 223 ARG   ( 409-)  A      N
 223 ARG   ( 409-)  A      NE
 255 ASN   ( 441-)  A      N
 255 ASN   ( 441-)  A      ND2

Warning: Unusual water packing

We implemented the ion valence determination method of Brown and Wu [REF] similar to Nayal and Di Cera [REF] and Mueller, Koepke and Sheldrick [REF]. It must be stated that the validation of ions in PDB files is very difficult. Ideal ion-ligand distances often differ no more than 0.1 Angstrom, and in a 2.0 Angstrom resolution structure 0.1 Angstrom is not very much. Nayal and Di Cera showed that this method nevertheless has great potential for detecting water molecules that actually should be metal ions. The method has not been extensively validated, though. Part of our implementation (comparing waters with multiple ion types) is even fully new and despite that we see it work well in the few cases that are trivial, we must emphasize that this method is untested.

The score listed is the valency score. This number should be close to (preferably a bit above) 1.0 for the suggested ion to be a likely alternative for the water molecule. Ions listed in brackets are good alternate choices. *1 indicates that the suggested ion-type has been observed elsewhere in the PDB file too. *2 indicates that the suggested ion-type has been observed in the REMARK 280 cards of the PDB file. Ion-B and ION-B indicate that the B-factor of this water is high, or very high, respectively. H2O-B indicates that the B-factors of atoms that surround this water/ion are suspicious. See: swift.cmbi.ru.nl/teach/theory/ for a detailed explanation.

 288 HOH   ( 110 )  A      O  0.88  K  4 ION-B
 288 HOH   ( 121 )  A      O  0.89  K  4
 288 HOH   ( 144 )  A      O  1.04  K  4 ION-B

Final summary

Note: Summary report for users of a structure

This is an overall summary of the quality of the structure as compared with current reliable structures. This summary is most useful for biologists seeking a good structure to use for modelling calculations.

The second part of the table mostly gives an impression of how well the model conforms to common refinement restraint values. The first part of the table shows a number of global quality indicators.


Structure Z-scores, positive is better than average:

  1st generation packing quality :   0.149
  2nd generation packing quality :  -0.372
  Ramachandran plot appearance   :   1.014
  chi-1/chi-2 rotamer normality  :   0.812
  Backbone conformation          :   0.284

RMS Z-scores, should be close to 1.0:
  Bond lengths                   :   0.384 (tight)
  Bond angles                    :   0.575 (tight)
  Omega angle restraints         :   1.030
  Side chain planarity           :   0.491 (tight)
  Improper dihedral distribution :   0.554
  B-factor distribution          :   0.367
  Inside/Outside distribution    :   0.939

Note: Summary report for depositors of a structure

This is an overall summary of the quality of the X-ray structure as compared with structures solved at similar resolutions. This summary can be useful for a crystallographer to see if the structure makes the best possible use of the data. Warning. This table works well for structures solved in the resolution range of the structures in the WHAT IF database, which is presently (summer 2008) mainly 1.1 - 1.3 Angstrom. The further the resolution of your file deviates from this range the more meaningless this table becomes.

The second part of the table mostly gives an impression of how well the model conforms to common refinement restraint values. The first part of the table shows a number of global quality indicators, which have been calibrated against structures of similar resolution.

Resolution found in PDB file : 1.42


Structure Z-scores, positive is better than average:

  1st generation packing quality :   0.3
  2nd generation packing quality :  -0.8
  Ramachandran plot appearance   :   0.4
  chi-1/chi-2 rotamer normality  :   0.3
  Backbone conformation          :  -0.2

RMS Z-scores, should be close to 1.0:
  Bond lengths                   :   0.384 (tight)
  Bond angles                    :   0.575 (tight)
  Omega angle restraints         :   1.030
  Side chain planarity           :   0.491 (tight)
  Improper dihedral distribution :   0.554
  B-factor distribution          :   0.367
  Inside/Outside distribution    :   0.939
==============

WHAT IF
    G.Vriend,
      WHAT IF: a molecular modelling and drug design program,
    J. Mol. Graph. 8, 52--56 (1990).

WHAT_CHECK (verification routines from WHAT IF)
    R.W.W.Hooft, G.Vriend, C.Sander and E.E.Abola,
      Errors in protein structures
    Nature 381, 272 (1996).
    (see also http://swift.cmbi.ru.nl/gv/whatcheck for a course and extra inform

Bond lengths and angles, protein residues
    R.Engh and R.Huber,
      Accurate bond and angle parameters for X-ray protein structure
      refinement,
    Acta Crystallogr. A47, 392--400 (1991).

Bond lengths and angles, DNA/RNA
    G.Parkinson, J.Voitechovsky, L.Clowney, A.T.Bruenger and H.Berman,
      New parameters for the refinement of nucleic acid-containing structures
    Acta Crystallogr. D52, 57--64 (1996).

DSSP
    W.Kabsch and C.Sander,
      Dictionary of protein secondary structure: pattern
      recognition of hydrogen bond and geometrical features
    Biopolymers 22, 2577--2637 (1983).

Hydrogen bond networks
    R.W.W.Hooft, C.Sander and G.Vriend,
      Positioning hydrogen atoms by optimizing hydrogen bond networks in
      protein structures
    PROTEINS, 26, 363--376 (1996).

Matthews' Coefficient
    B.W.Matthews
      Solvent content of Protein Crystals
    J. Mol. Biol. 33, 491--497 (1968).

Protein side chain planarity
    R.W.W. Hooft, C. Sander and G. Vriend,
      Verification of protein structures: side-chain planarity
    J. Appl. Cryst. 29, 714--716 (1996).

Puckering parameters
    D.Cremer and J.A.Pople,
      A general definition of ring puckering coordinates
    J. Am. Chem. Soc. 97, 1354--1358 (1975).

Quality Control
    G.Vriend and C.Sander,
      Quality control of protein models: directional atomic
      contact analysis,
    J. Appl. Cryst. 26, 47--60 (1993).

Ramachandran plot
    G.N.Ramachandran, C.Ramakrishnan and V.Sasisekharan,
      Stereochemistry of Polypeptide Chain Conformations
    J. Mol. Biol. 7, 95--99 (1963).

Symmetry Checks
    R.W.W.Hooft, C.Sander and G.Vriend,
      Reconstruction of symmetry related molecules from protein
      data bank (PDB) files
    J. Appl. Cryst. 27, 1006--1009 (1994).

Ion Checks
    I.D.Brown and K.K.Wu,
      Empirical Parameters for Calculating Cation-Oxygen Bond Valences
    Acta Cryst. B32, 1957--1959 (1975).

    M.Nayal and E.Di Cera,
      Valence Screening of Water in Protein Crystals Reveals Potential Na+
      Binding Sites
    J.Mol.Biol. 256 228--234 (1996).

    P.Mueller, S.Koepke and G.M.Sheldrick,
      Is the bond-valence method able to identify metal atoms in protein
      structures?
    Acta Cryst. D 59 32--37 (2003).

Checking checks
    K.Wilson, C.Sander, R.W.W.Hooft, G.Vriend, et al.
      Who checks the checkers
    J.Mol.Biol. (1998) 276,417-436.