Volume 1 Chapter 7 3D Geometrical Surveys as Precursors to 3D Searching

Back to Table of Contents

7.10 Exhaustive Fragment Location

A 3D geometric search also differs significantly from a 2D chemical search in terms of the basic mechanism of fragment location in the two cases, i.e. in the matching of the query substructure to the connectivity records held in the CSD.

Most 2D searches are designed to answer the question: "how many entries in the CSD contain this chemical substructure?" Hence, the test is satisfied as soon as a single instance of the query substructure is located in a database entry. On a very few occasions we might wish to locate CSD entries that contain a specified minimum number of occurrences of a given substructure, and the 2D-CONSTRAIN instruction 'NUMBER-OF-FRAGS' is provided for this purpose.

Thus, in the 2D chemical substructure search, each and every fragment located is identical both to the query and to every other located fragment.

In the 3D case, the situation is very different. Here we must locate all unique examples of the 3D crystal fragment in each entry and, for each one, proceed to calculate the required geometrical parameters and apply the 3D constraints.

This need for exhaustive fragment location is due to the fact that, although the 2D chemical structure is identical in every case, the 3D structure is almost certainly different. Some crystal fragments may pass the 3D search criteria and be classified as hits; others will not.

For example, we may search for boat-form six-membered rings by encoding appropriate constraints on the six intra-annular torsion angles. In a given CSD entry, we may locate a number of examples of six-membered rings, but perhaps only one of these will satisfy the 3D constraints. However, in order to carry out the 3D search operation systematically, each and every example of the 6-membered ring substructure which occurs in the CSD entry must be located and tested.

In crystal structures there are more problems to consider in carrying out the exhaustive fragment location.

First, we must consider molecules having internal symmetry and where the molecular symmetry coincides with an appropriate symmetry element in the crystal. In these cases, the molecule stored in the CSD and on which the search operates will consist of a number of unique atoms (the asymmetric unit), together with another set (or sets) of 'symmetry-related' atoms bonded to the atoms of the asymmetric unit to form the complete chemical molecule.

In these cases substructural fragments located from the symmetry-related set of atoms and the geometry derived therefrom will be identical to that derived from fragments located in the asymmetric unit.

Secondly, the asymmetric unit of a crystal structure can contain more than one example of the chemical molecule stored in the CSD. Thus we may locate only one example of a substructure in the chemical connection table, but two, three or, on rare occasions, more than three unique examples of the crystal fragment may exist in the crystal structure.

This situation is also accommodated in the 3D search routines of QUEST3D. This is necessary, since the geometry of each fragment may vary between the different molecules that comprise the crystal asymmetric unit. In some cases this variation may be large enough for one or more of the fragments to fail the 3D search criteria, but for others to be classified as hits.

For these reasons, the default and safest mode of operation of the 3D search routines is exhaustive fragment location. It may be that, in some special cases, the user will wish to alter this default setting.


Back to Table of Contents

Volume 1 Chapter 7 Topological Symmetry and Fragment Location.