Volume 1 Chapter 7 Display of Hits from Non-Bonded Searches
Back to Table of Contents
A complete set of subfragments, linked together by non-bonded contacts, once
located, can be treated in the same way as a bonded fragment. Thus, we may use
the 3D-CONSTRAIN sub-menu to SETUP geometric objects, DEFINE and name
geometrical parameters, TRANSFORM those parameters, and SELECT some subset of
hits on the basis of geometrical parameter values.
The calculation of such geometric data is common in the systematic study of
hydrogen bond interactions.
Ex. 11 We extend the study of amino-thiocarbonyl hydrogen bonding begun in
Ex. 7 of section 7.14.3 to:
The setup of the search in the 3D-CONSTRAIN sub-menu and an example hit, and
the summary statistics of all hits are given are shown below.
NFRAG REFCOD HBOND HANG THETA PHI
1 ABHYTZ 2.480 164.762 -20.906-106.093
2 ACRTUR 2.589 157.488 32.379 106.570
3 ACRTUR 2.561 170.452 -7.296-105.989
4 ACTCBZ 2.931 137.567 46.311 121.393
685 XANHYD 2.112 160.436 17.513 101.275
686 XANHYD01 2.138 160.395 -18.660 100.561
687 ZNDITZ02 2.856 124.215 34.757-151.824
688 ZZZEII01 2.890 157.984 -54.200 160.643
689 ZZZGEO01 2.546 169.119 -6.861-109.372
690 ZZZMJK02 2.634 155.342 -17.372 108.877
691 ZZZQMK10 2.520 140.496 -71.513 132.202
Nobs 691 691 691 691
Mean 2.604 156.226 -2.651 -5.224
SDSample 0.176 14.416 38.270 111.815
SDMean 0.007 0.548 1.456 4.254
Minimum 2.058 120.578 -85.672-179.076
Maximum 2.997 179.169 85.205 179.849
It is obvious that the intermolecular non-bonded search mechanism is highly
cpu-intensive, since it can involve the testing of an appreciable number of
symmetry/translational variants of atomic positions. Even though the
symmetry/translation search routine employs upper-bound tests and is as
efficient as possible, the intermolecular searches are likely to take much
longer than any comparable intramolecular search.
In the present database and software release, no 3D screening system exists to
speed these intermolecular searches. The only screening procedures employed
relate to the 2D chemical structures of the subfragments that make up the
complete non-bonded search query. This lack of 3D intermolecular distance
screens is currently being remedied.
For the present, several factors should be considered when composing non-bonded
Back to Table of Contents
- Choose distance criteria carefully. Remember that the search space is
proportional to the third power of the distance.
- The larger the 2D subfragments, the greater the possibility for generating
a reasonable number of 2D screens.
- Incorporate user-defined screens whenever possible. For example, restrict
the search to entries that are error-free, not disordered, have R-factors below
some cut-off value, or are drawn from just the organic (or metallo-organic)
sections of the database only.
- CONTACTs that use GROUPs are searched more quickly by centroid
representations than by atom-by-atom methods.
- Exercise care when conducting searches with common functional groups, e.g.
C=O, N-H, and aromatic rings. Conduct a quick 2D search to determine the number
of entries that contain the subfragments that will be used in the nonbonded
search. If the number is large consider limiting the non-bonded search to a
particular class of entries as a first pass.
- Subfragments that have a high degree of topological symmetry (e.g. rings)
will take longer in exhaustive searches.
Volume 1 Chapter 7 Further Examples of Non-Bonded Searches.