LRA, Vol. 12, pp.
37–40
On the DZKP Molecular
Descriptors
D. Zakarya,1 M. Nohair,1 and H. Nyassi2
1
Faculty of Sciences and Technology, B.P. 146, 20650 Mohammedia, Morocco, Tel: (00-
212) (3) 31-47-05/31-47-08; Fax: (00-212) (3) 31-53-53; E-mail: Zakaryadriss@hotmail.com
2
Faculty of Sciences, El Jadida, Morocco
Received 14 October 1999
ABSTRACT: In this article, we give some information
about the computation and the selectivity of DZkp de-
scriptors for a set of chemical structures. The rela-
tions between the descriptors, autocorrelation com-
ponents, and Wiener’s index are given. q 2000 John
Wiley & Sons, Inc. Lab Robotics and Automation
12:37–40, 2000
INTRODUCTION
Establishment of structure-activity relationships be-
comes actually a very interesting research field. Elab-
oration of such relations requires a set of molecules
and their activity or property. It is actually admitted
that the activity or the property of the molecule gen-
erally depends on its chemical structure [1]. Thus the
reliability of structure-activity relationships highly
depends on the chosen parameters to account for the
chemical structure.
Numerous methods and approaches were intro-
duced to describe the chemical structure for a given
set of molecules and several of them are based on the
Correspondence to: D. Zakarya.
q 2000 John Wiley & Sons, Inc.
molecular graph in conjunction with some chemical
or physico-chemical properties. Among the large va-
riety of descriptors, topological indices (TIs) are
based on the graph of the molecule in conjunction
with the topological aspect of the atoms or groups.
Although TIs were used successfully in many studies
[2–4], several of them can not describe the chemical
structure with high chromatism. We think that is gen-
erally a reason that TI presents low selectivity. This
drawback was avoided by introduction of new top-
ological indices like MTI [4]. Elaboration of the new
descriptors called DZkp [5] was based on a combina-
tion between the concept of the original Wienner’s
index [6] and the concept of chromatism developed
in the autocorrelation method [7] and DARC system
[8].
We will give in this communication the general
procedure for the calculation of such descriptors, and
the selectivity and chemical information that some
DZkp descriptors account for.
DZkp DESCRIPTORS
The first reported use of a topological index in chem-
istry was by Wienner. He used it for the study of the
structure and properties of standard hydrocarbons.
37
38 Zakarya, Nohair, and Nyassi

This index is defined as the half-sum of elements of ● k is the distance considered for the calculation
the distance matrix. of DZkp descriptors (see examples)
W 4 0.5 od ij (1) ● dij is equal to the length of the shortest path
between vertices i and j
● dij is the element of the distance matrix for the
● P(i) and P(j) are properties or contribution to
graph G of the molecule
properties (e.g., van der Waals volume or sur-
● dij is equal to the length of the shortest path face, electronegativity, or molar refraction) of
between vertices i and j atoms i and j in the molecule. The hydrogen
The autocorrelation components are also based on atoms are not taken into account. On the basis
the distance matrix in conjunction with properties of our previous works concerning the calcula-
(or contributions of properties) of the atoms of the tion of the autocorrelation components [9], we
molecule (vertices of the graph G). generalize the equation (3), which becomes.
Cdij 4 o f(i)f(j) (2) k

Our idea consists in a combination of the concept of

DZkP 4 o d o P(i)
1
ij
dij j.i
x
P(j)x (4)
calculation of W and the concept of the calculation
of the autocorrelation components. This leads to the ● x 4 1 (like 1 in autocorrelation method [7]) or
DZkp descriptors, calculated as follows: 0.5 (like in modified autocorrelation method
k
[10])
DZpk 4 o d o P(i) P(j)
1
ij
dij j.i
(3),dij 4 dk DZkp is a summation of autocorrelation components
multiplied by the distance for which the component
is calculated.
TABLE 1. DZkp Values for Some Chemical k

Structures DZkp 4 o
dij41
dij Cp (5)
Molecule DZ2v DZs2 DZc2
Cp is the component of autocorrelation method for a
CH2-CH2-CH3 50.99 7.62 4.83
given property (e.g. van der Waals volume [10])
CH3-CH2-OH 41.87 6.44 4.83
CH3-CH2-Br 52.02 7.57 4.83
CH3-CH2-Cl 47.94 7.16 4.83 Example of the Computation of DZkp
The following is an example of the computation of
DZkp:
TABLE 2. DKkp Values for Alkanes C2–C11a
● Molecule: propane (CH3CH2CH3)
Nc DZ1c DZc2 DZc2 DZc4 DZc5
● P 4 connectivity (number of neighbors of the
2 1 1 1 1 1
atom) and x 4 0.5
3 2.828 4.828 4.828 4.828 4.828
4 4.828 10.484 13.484 13.484 13.484 ● x 4 0.5
5 6.828 16.484 24.968 13.484 13.484
For k 4 1, DZ1c 4 (k 4 1) [(10.5 2 20.5) `
6 8.828 22.484 36.968 48.280 53.280
7 10.828 28.484 48.968 68.280 82.420 (1 2 20.5)] 4 2.828
0.5

8 12.828 34.484 60.968 88.280 12.420 This corresponds to the two pairs of atoms with a
9 14.828 40.484 72.968 108.28 142.420 distance of 1 bond and connectivities respectively
10 16.828 46.484 84.968 128.280 172.420 equal to 1 and 2.
11 18.828 52.484 96.968 148.280 202.420 For k 4 2, DZ2c 4 1 (k) [(10.5 2 20.5) ` (10.5 2
a
x 4 0.5. 2 )] ` 2(k) (10.5 2 10.5) 4 4.828
0.5

TABLE 3. Correlation Coefficient (r) and Standard

SELECTIVITY OF DZkp
Deviation (s) Values for the b.p. 4 f(DZkp)
Relationships According to the equation (4), it is possible to derive
k 1 2 3 4 5 Nc a nonfinite number of molecular descriptors depend-
ing on k and p. For example, we choose four mole-
r 0.89 0.82 0.89 0.95 0.95 0.95 cules, having the same graph, to illustrate calculation
S 11.5 14.4 11.7 8.2 8.2 7.20
of some DZkp descriptors (Table 1). Here p is the van

Figure 1. DZkp (p 4 connectivity) vs. number of carbon atom (Nc).
der Waals volume and surface taken from reference
[10], connectivity (number of neighbors of the atom).
The considered graph is hydrogen-depleted, x 4 0.5,
and k 4 2.
Table 1 shows that the connectivity is not a dis-
criminant factor between the molecules studied, in
opposite to the other properties. The selectivity [11]
(number of different values (50.01)/number of total
compounds) of DZkp (p, connectivity; x 4 0.5) was
tested for a set of 300 alkanes [10] (C6–C11). We found
that, it increases with the increase of k and it is suf-
ficiently high (95%) for k 4 3.
In the following, we examine the variation of
DZkp descriptors (p, connectivity) with the size of the
molecule for a set of linear alkanes (C2–C11). The size
is expressed here by the number of carbon atoms (Nc)
(Table 2 and Figure 1).
Figure 1 shows that the linearity of DZkp descrip-
tors with Nc depends on k. This graph may be useful
when DZkp descriptors are chosen for structure-prop-
erty (activity) relationship studies. For example, it is
well known that boiling point (b.p.) is a nonlinear
function of Nc. Then we must find that DZkp corre-
sponding to lower k values is not well correlated with
the b.p. To illustrate this hypothesis, we attempted
to correlate DZkp descriptors (k 4 1 to 8; and p, con-
nectivity) with the boiling points for a set of 66 al-
kanes (C6–C9). The results are given in Table 3.
These results show that the adequate k value is
4 or 5. An equivalent correlation was obtained (r 4
0.95, s 4 8) when an extended form (eDZpk ) of DZpk
descriptors was used.
k
e
DZkp 4 o
i
f(i) ` o d o f(i)
1
k
dij j.i
x
f(j)x
It is not surprising that the Nc leads to comparative
On the DZ PK Molecular Descriptors 39
results to that obtained with the others descriptors.
However, it is a nonselective variable between iso-
mers and does not certainly lead to good results for
many structure-activity relationships.
The results given here are less interesting than
several ones published in the literature. However,
our aim is only the illustration of the use of DZkp as
possible descriptors in QSARs and similar fields.
Efficiency of DZkp descriptors is not well demon-
strated here in comparison with Wiener’s index.
However, we have shown in a previous work [6] that
DZkp descriptors were more adequate in the chemical
structure description.
REFERENCES
[1] Needham, E. D.; Chien Wei, I.; Seybold, P. G. Molec-
ular modeling of the properties of alkanes. J Am
Chem Soc 1988, 110, 4186–4194.
[2] Seybold, P. G.; May, M.; Bagal, U. A. Molecular struc-
ture-property relationships. J Chem Edu 1987, 64,
575–581.
[3] Balaban, A. T. Applications of graph theory in chem-
istry. J Chem Inf Comput Sci 1985, 25, 334–343.
[4] Schultz, H. P. Topological organic chemistry, I:
Graph theory and topological indices of alkanes. J
Chem Inf Comput Sci 1989, 29, 227–228.
[5] Zakarya, D. Utilisation des composantes
d’autocorrélatio et de l’indice de Wienner dans
l’évaluation du sueuil olfactif d’alcools aliphatiques,
(6) New J Chem 1992, 16, 1039–1042.
[6] Wiener, H. Structural determination of paraffin boil-
ing points. J Am Chem Soc 1947, 69, 17–20.
40 Zakarya, Nohair, and Nyassi
[7] Moreau, G.; Broto, P. Nouv J Chim 1980, 4, 757–764.
[8] Dubois, J. E.; Mercier, C.; Panaye, A. DARC topolog-
ical system and computer aided design. Acta Pharm
Yougosl 1986, 36, 135–169.
[9] Zakarya, D.; Chastrette, M.; Tiyal, F. Use of the mul-
tifunctional autocorrelation method to estimate mo-
lar volumes of alkanes and oxygenated compounds.
J Org Phys Chem 1993, 6, 574–582.
[10] Tiyal, F. Ph.D. Thesis, University of Lyon I, Lyon,
1990.
[11] Razinger, M.; Chrétien, J. R.; Dubois, J. E. Structural
selectivity of topological codes in alkanes series. J
Chem Inf Comput Sci 1985, 25, 23–27.