2D 3 Point Descriptors

2D 3 Point Descriptors

The 2D 3 Point Descriptors are build from the 2 dimensional structure of the compounds, this means that distances are not real space distances from a single conformation but number of bonds between two atoms in the molecule. The 3 Points in the descriptor name refers to the 3 points which are taken together to build one feature. Lets have a look at one example:

In this example one of many features, the "HA 4 HA 2 Hal 4" feature is shown. Which consist of 2 hydrogen bond acceptors (HA) and one halogen atom, which have distances of 4, 2 and 4 bonds from each other in this molecule. The points which are used to define the corners of this triples are called pharmacophore point.The following points are corrently used:
NameMeaningSMARTS
HAHydrogen Acceptor[$([$([#8,#16]);!$(*=N~O);!$(*~N=O);X1,X2]),
$([#7;v3;!$([nH]);!$(*(-a)-a)])]
HDHydrogen Donor[$([O;H1,-&!$(*-N=O)]),
$([S;H1&X2,-&X1]),
$([#7;H;!$(*(S(=O)=O)C(F)(F)F);!$(n1nnnc1);!$(n1nncn1)]),
$([#7;-])]
AcAcidic Group[$([O;H1]-[C,S,P]=O),
$([*;-;!$(*~[*;+])]),
$([NH](S(=O)=O)C(F)(F)F),
$(n1nnnc1),
$(n1nncn1)]
BaBasic Group[$([NH2]-[CX4]),
$([NH](-[CX4])-[CX4]),
$(N(-[CX4])(-[CX4])-[CX4]),
$([*;+;!$(*~[*;-])])$(N=C-N),
$(N-C=N)]
RAAliphatic Ring Attachment[$([A;D3](@*)(@*)~*)]
AAAromatic Ring Attachment[$([a;D3](@*)(@*)*)]
HPHydrophobic Group[C; !$(C-[!C]); !$(C-C-[!C]); !$(C=,#*); !$(C-C=,#*)]
HalHalogen[F,Cl,Br]
XNon Organic Atom[!#1;!#6;!#7;!#8;!#9;!#16;!#17;!#35]

Using 2D descriptors has advantages and disadvantages:

Within WATER large lists of this 2D 3 Point pharmacophores are computed for each molecule. Such a list called pharmacophore key contains around 3000 features from a possible list of about 300 000.

To improve the model, keys representing multiple occurences of a feature are also used. You may recognice them by a suffix like "C=2" which means that this feature occured at least twice in this compound.

To further increase flexibility there is some fuzzyness added to the distances. The "HA 4 HA 2 Hal 4" key is also recorred as:
HA -103 HA 2 Hal 4
HA -104 HA 2 Hal 4
HA -105 HA 2 Hal 4
HA 4 HA -101 Hal 4
HA 4 HA -102 Hal 4
HA 4 HA -103 Hal 4
HA 4 HA 2 Hal -103
HA 4 HA 2 Hal -104
HA 4 HA 2 Hal -105
This allows you to find cases where the distance of eg. the two HA may be 3 or 4 for a compound to be active.


Page Owner: Alberto Gobbi - Last updated: Apr 17, 1999