Volume 1 Chapter 7 Selecting Suitable 3D Search Constraints

Back to Table of Contents

7.9 3D Geometrical Surveys as Precursors to 3D Searching

The facilities of QUEST3D and the associated program GSTAT now provide the ability to carry out thorough surveys of fragment geometry as an aid to establishing appropriate 3D geometrical search limits.

Let us return to the morphine example introduced at the beginning of this chapter:

SETUP X1  7  8  9  10  11  12
SETUP P1  7  8  9  10  11  12
SETUP P2  1  2  3  4  5  6
DEFINE NANG  1  6  5
DEFINE N - C1  1  6
DEFINE N - C2  5  6
DEFINE DIST  X1  6
DEFINE ?ANG  P1  P2
TRANSFORM ?SIN = SIN  ?ANG
TRANSFORM PANG = ASIN  ?SIN
DEFINE ?DEV  6  P1
TRANSFORM NDEV = ABS  ?DEV
SELECT DIST  4.45  4.75

As coded, with limits of 4.45 to 4.75Å for DIST, the search is a success. However, if we look at the histogram of DIST for the hits,

Histogram of DIST

 
DIST        2    4    6    8   10   12   14   16   18   20
       .....I....I....I....I....I....I....I....I....I.....
  4.45 -**                                               - (  1)
       .**************************                       . ( 13)
       .**************                                   . (  7)
       .**************                                   . (  7)
       .********                                         . (  4)
  4.52 -******                                           - (  3)
       .****                                             . (  2)
       .**************                                   . (  7)
       .********                                         . (  4)
       .********                                         . (  4)
  4.60 -******                                           - (  3)
       .********                                         . (  4)
       .**                                               . (  1)
       .**                                               . (  1)
       .                                                 .
  4.67 -                                                 -
       .                                                 .
       .                                                 .
       .                                                 .
       .                                                 .
  4.75 -                                                 -
       .....I....I....I....I....I....I....I....I....I.....
DIST        2    4    6    8   10   12   14   16   18   20

we note that:

Furthermore, the histogram tells us nothing about the 32 instances of the fragment that were eliminated by the 3D search constraint:
 SELECT  DIST  4.45  4.75

A more informative approach would have been to examine histograms for DIST, and a variety of other parameters, over all instances of the fragment. Scattergrams are invaluable in plotting the variance of one parameter against a second parameter.

In choosing the explicit geometrical parameters for this fragment we have also chosen to calculate:

PANG, the acute angle between the mean planes through the aromatic and heterocyclic rings

NDEV, the deviation (in Å) of nitrogen from the mean plane defined by the aromatic ring atoms.

The scattergrams of DIST versus both PANG and NDEV are shown below.

Scattergram of DIST versus PANG

Scattergram of DIST versus NDEV

Now, one of the reasons for chosing DIST as a parameter that is characteristic of this fragment is an assumption that the bridging in the fragment will drastically reduce its conformational flexibility. However, both of the scattergrams (particularly the DIST - PANG plot) indicate that the fragment is more flexible than might have been imagined. In fact all but one of the 93 heterocyclic rings adopts a chair conformation (determined by examining the relevant torsion angles), but the ring can rotate by up to 20deg. from an obviously preferred position which is perpendicular to the aromatic ring.

It can be argued that the almost linear relationship observed in both scattergrams must arise from structural constraints in the fragment itself. However, this is not the primary point of the data visualisation afforded by the scattergrams. What is important is the position of parent morphine - not only does morphine occur at the high end of the DIST distribution but it also shows one of the highest degrees of heterocyclic ring rotation away from being perpendicular to the aromatic ring (PANG = 78deg.) and one of the closest nitrogen-to-aromatic plane distances (NDEV = 0.68Å).

These scattergrams show that DIST alone is not, perhaps, sufficiently characteristic of the fragment. To answer the real query: "which fragments in CSD have geometries that are most similar to parent morphine?", then a 3D search based on DIST and one of PANG or NDEV might well be regarded as a significant improvement. Appropriate limiting values can be read from the scattergrams.

Back to Table of Contents

Volume 1 Chapter 7 Exhaustive Fragment Location.