Indian Journal of Chemistry Vol. 45A, January 2006, pp. l t 1-125

Application of quantUlll chemical descriptors in computational medicinal chemistry and chemoinformatics

R Parthasarathi", M Elango", J Padmanabhan", V Subramanian a.*, D R Roy b, U Sarkarb & P K Chattaraj"'*

"Chemica! Laboratory, Central Leather Research Institute, Adyar Chennai 600 020, India Email: subuchem@hotmail.com

b Department of Chemistry, Indian Institute of Technology, Kharagpur 72l 302, India Email: pkc@chem.iitkgp.emet.in

Received 17 November 2004; revised I J November 2005

During the practice of chemoinformatics, it has been realized that molecu!ar diversity is an essential feature to characterize the reactivity of the molecules. In addition, it is of utmost importance to enrich potential libraries with those molecules which could be converted to suitable drug candidates or omited as toxins. In addition, a paradigm shift in structure-activity relationship has resulted in the integration of various descriptors and quantum chemical descriptorS based drug development activities into early stages of lead discovery. In particular, various descriptors are being developed and used to help identify and screen out compounds that are unlikely to become drugs/toxins. This paper highlights the developmcnt of recent Density Functional Theory (DFT) based chemical reactivity descriptors and the applications of these descriptors towards the prediction of chemical reactivity, especially in the prediction of toxicity, biological activities and other chemicai informatic properties as well as reactive site and group identification and recent developments towards recogniticn of potentially toxic molecules. Bridging experimental knowledge with effective computational infoimation, managemellt and prediction of various aspects of molecular reactivities thus facilitates the rapid and cost-effective processes and help~ focus attention on interesting mo.lecules.

The use of i!lformation technology and management has become :l critical part of the drug discovery process. The rational design of new drug molecules involves inpL:t from various branches of. soience. in this context ine information and management of bio and chemica! information have become the integral part. The data base development, management and analysis of biological information are defined as bioinformatic,; 1,2 which includes various database

science, and cheminformatics are some of the related areas of chemoinformatics. The development of future chemical informatics systems will require a workforce with a solid grounding in chemistry and an expert understanding of the available computer technology. Chemical, agrochemical, pharmaceutical, and biotechnology branches of science require extensive input from both bioinformatics and chernoinfonnatics.

management tools, analysis tools and molecular The quantitative structure-acttvlty relationship modeling. The term "chemoinformatics" has been (QSAR) and the quantitative stl;1.lclUre-property introduced in the Annual Reports of Medicinal relationship (QSPR) are the irnporta~£ topis . of the Chemistry in 1998 by Brown. Chemoinformatics l

-IO bio-chemo-informatics which can ~ebuili essentially

is the ama1samation of those chemical information based 00 the data generated from the molecular resources to transform data into vital information and modeling and computational chemistry_ The QSAR chemical information into knowledge for the intended and QSPR attempt to find a mathematical relationship purpose of making better decisions faster in the area between chemical structure and biological . activity or of drug lead identification and organization. In fact, chemical property for a series of homologous both bioinforrnatics and chemoinformatics are generic compounds. These series of homQlogous comp()!Jnds terms that enco.mpassthe design, creation, 'are called the training set, T~egenei:ated organization, miulagement, retrieval, analyst~, ' mathematical equation can be used to predict the disseminatior:, vistHi.llzation arid use of chemical drid activity or property of any new.compotin9,whiGh has brologicaJ information . . :. ': :t:hemi~informatiGs, peen built from the chosen traifling:"$~L,Nu~rous chemornciric::, computational " 'chemistry, chemicaJ . ,;.,d~~~riptors have beeoused to ,develpPnQSAR and } r; fo ~rr1'JJ; (;S chemical infor~~tibn management! : , ;QSPR for different applications. In 'thi~;}j~gard it is

112 INDIAN 1 CHEM. SEC A. JANUARY 2006

necessary to mention the noteworthy contribution made by Hansch and coworkers2

-5 to the development

and growth of this area of activity_

Quantum chemical methods and molecular modeling techniques enable the definition of a large number of molecular and local quantlhes characterizing the reactivity, shape and binding properties like atomic charges, molecular orbital energies, frontier orbital densities, superdelocalizabilities, atom-atom polarizabilities, molecular polarizability, dipole moment and polarity indices and energy of a complete molecule as well as of molecular fragments and substituents_ The study of a structure-activity relationship (SAR) for a series of compounds showing some specific bioactivity has been investigated by Kobayashi et aL II This study has focused on the understanding of the interaction between a drug and an enzyme in the actual drug­enzyme complex using its "intrinsic" reactivity, expressed in terms of absolute hardness and absolute electronegativity _ Semiempirical calculations were used in combination with the calculations of the hardness and electronegativity values for a series of quinolines, 1,8-naphtheridines, and polychlorinated dibenzo-p-dioxins and biphenyls and a study has been carried out to establish the relationship between antibacterial activity with their respective Off descriptors_ The dioxins and biphenyls have shown activity as xenobiotics by means of noncovalent binding to the arylbydrocarbon receptor (AhR)12·13. In establishing the SAR, the property-actlVlty relationship (PAR), the TJ-X activity diagrams have been used to decipher the correlation_ It is evident from this study that the compounds showing high xenobiotic activity are found to be hard, a trend perfectly in accordance with the fact that a noncovalent, electrostatic interaction is involved in the mode of action. In the case of the antibacterials, the hardness could not discriminate between active and nonactive compounds. However, the electronegativity could provide a better picture. In contrast to the case of the hardness-controlled activity, electrostatic interactions are not predominant here_ This clearly indicated that the process of charge transfer between the two interacting compounds plays an important role. As the amount of charge transfer is partially determined by the electronegativity, the importance of this descriptor in this case study can thus be rationalized. The electronegativity is not the only quantity that plays a role in this context as

revealed in the recent study by Maynard and Covell 14.

They have used a new Off descriptor namely. electrophilicity to establish the SAR in the development of anti-human immunodeficiency virus (anti-HIY) drugs, which is extremely important because of the emergence of protease and reverse transcriptase drug-resistant HIV strains. An excellent linear correlation of the electrophilicity with the logarithm of the observed reaction rate constants has been found. This investigation has also led to the formal definition of the electrophilicity by Parret al. 15

The Off offers a strong foundation for various qualitative concepts in the chemical reactivity. Popular qualitative chemical concepts such as electronegativity and hardness have been widely used in . understanding various aspects of chemical reactivity I6-21. Recently, Geerlings and co-workers have reviewed the tremendous development in the application of conceptual density functional theoryl? to variety of chemical and biological problems. The nature of basic chemical concepts, electronegati vity, hardness and softness, called Global Reactivity Descriptors (GRD), has been theoretically justified within the framework of Off. Along with these global descriptors, other important Local Reactivity Descriptors (LRD), such as Fukui function and local softness were also proposed to rationalize the reactivity of a particular site in a molecule. A series of new reactivity descriptors such as electrophilicity, local electrophilicity index and group philicity have been defined to understand the chemical reactivity and site selectivity. Generally, GRD are used to probe the global reactivity of the molecules whereas the LRD provide information about the particular site in the molecule. Both GRD and LRD have been used in numerous occasions to probe the chemical reactivity and site selectivity in static and dynamic situations involving ground and excited electronic states22

-31

•

These reactivity descriptors are better appreciated in terms of associated electronic structure principles such as electronegativity equalization principle (EEP). hard-soft-acid-base (HSAB) principle, maximum hardness principle (MHP) and minimum polarizability principle (MPP). The formal definitions of all these descriptors and their applications have been presented . al I . 16 17 H ., be 10 sever e egant reViews '. owever, lor tter understanding and clarity, the definitions of all these descriptors are presented in the paper.

PARTHASARATHI et al.: QUANTUM CHEMICAL DESCRIPTORS IN CHEMOINFORMATICS 113

Theoretical Background Reactivity descriptors

The quantitative definitions for chemical potential (Il) and electronegativity (X)32 for an N-electron system with total energy E can respectively be given as:

[aE] Jl.= -aN v(i')

... (1)

and

... (2)

where vCr) is the external potential. Chemical hardness (10 has been identified as an

useful global reactivity index in atoms, molecules and clusters32.33. The theoretical defini tion of chemical hardness has been provided by Off as the second derivative of electronic energy with respect to the number of electrons N, for a constant external potential vCr)

.. . (3)

The corresponding global softness is expressed as:

... (4)

Using a finite difference method, working equations for the calculation of electronegativity and chemical hardness can be given by:

IP+EA X 2 ;11

IP-EA

2 ... (5)

where IP and EA are ionization potential and electron affinity of the atom or molecule respectively. If

E HOMO and E LUMO are the energies of the highest

occupied and lowest unoccupied molecular orbitals respectively, then the above equation can be rewritten using Koopmans ' theorem33 as:

X E HOMO +E LUMO

2 .. . (6) ; 11 E LUMO -E HOMO

2

Concept of Electrophilicity (W)1 5 has been introduced by Parr et al.. as a global reactivity index ~imilar to chemical hardness and chemical potential.

This new reactivity index measures the stabilization in energy when the system acquires an additional electronic charge L1N from the environment. The electrophilicity is defined'5 as :

.. . (7)

In Eq. (7), Il and TI are the electronic chemical potential and the chemical hardness of the ground state of atoms and molecules, respectively, approximated in terms of the vertical ionization potential I and electron affinity A . The electrophilicity is a functional descriptor of reactivity that allows a quantitative classification of the global electrophilic nature of a molecule within a relative scale.

The Fukui Function (FF)34 is one of the widely used local density functional descriptors to model chemical reactivity and site selectivity. The atom with the highest FF is highly reactive compared to the other atoms in the molecule. The FF is defined as the derivative of the electron density p (r) with respect to the total number of electrons N in the system, at constant external potential v ( r) acting on an electron due to all the nuclei in the system,

.. . (8)

where Jl. is the chemical potential of the system. It is more convenient to represent the FF values around each atomic site into a single value that characterizes the atoms in a molecule. Depending on the electron transfer, three types of FF35

-38 are defined as:

for nucleophilic attack

.. . (9a)

for electrophilic attack

.. . (9b)

f O (r)=[PN+1 (r)- PN - 1 (T»)/2 for radical attack

. .. - (9c)

The condensed FF are calculated using the procedure proposed by Yang and Mortier35 based on a finite difference method,

f/ =qk (N + 1)-qk (N) for nucleophilic attack

(lOa)

K =qk (N)-qk (N -1) for electrophilic attack

.. . (lOb)

114 INDIAN J CHEM, SEC A, JANUARY 2006

(IOc) where qk is the electronic population of atom k in a

molecule. Similar to FF, another local quantity called local

softness38 is defined as,

S(F)=(ap(r)] all v(i')

so that,

f s(r)dr=S

... (11)

... (12)

The above expression provides same information as FF except the additional knowledge of Sin s(r) and

has been widely used to describe the reactivity of atoms in molecule. Different local softnesses corresponding to nucleophilic, electrophilic and radical attacks can be defined as,

. . . (13)

where (cx= +, - and 0) represents local softness quantities describing nucleophilic, electrophilic and radical attacks, respectively. Based on local softness, relative nucleophilicity (sk-/se) and relative electrophilicity (sk+/sk-) indices have also been defined and their usefulness to predict reactive sites has also been addressed to. There are numerous studies on the applications of FF to model reactivity and site selectivity. It has been established that quantum chemical model selected to derive wave function, population scheme used to obtai~ the partial charges and basis set employed in the molecular orbital calculations are important parameters, which significantly influence the FF values. To understand the intermolecular reactivity, Pal and coworkers39

have used various local reactivity descriptors. It was found that the local softness could not provide correct intermolecular reactivity trend and hence they have suggested group softness as a local reactivity d . 39 Tl escnptor . le group softness is defined as :

1/

Sx = LSk k= l

. .. (14)

Recently, the generalized concept of philicity was proposed by Chattaraj et al.29

. It contains almost all information about hitherto known different global and

local reactivity and selectivity descriptors, in addition to the information regarding electrophilic/nucleophilic power of a given atomic site in a molecule. It is possible to define a local quantity called philicity associated with a site 'k' in a molecule with the aid of the corresponding condensed-to-atom variants of

Fukui function It as29,

Of=WFU k '1k .. . (15)

where (cx= +, - and 0) represents local philic quantities describing nucleophilic, electrophilic and radical attacks. Eq. (15) predicts that the most electrophilic site in a molecule is the one providing the maximum value of Ulk+' This site also coincides with the softest site in a molecule. When two molecules react, which one will act as an electrophile (nucleophile) will depend on one, which has a higher (lower) electrophilicity index. This global trend originates from the local behaviour of the molecules or precisely the atomic site(s) that is(are) prone to electrophilic (nucleophilic) attack. Chattaraj et al. established a generalized treatment of both global and local electrophilicities, as well _,s nucleophilicities. Numerous applications of this generalized philicity have been carried out. The toxicity of polychlorinated biphenyls (PCB), benzidine and dibenzofuran has been quantified with the help of unified philicity. The results revealed that generalized philicity provides better description of local reactivity along with other global reactivity descriptors. Effect of electric field on the global and local reactivity indices including philicity of various molecules has been analyzed. Recently, it has been shown4o that the philicity

. properly explains the intermolecular reactivity in most cases for a series of carbonyl compounds although the global electrophilicity fails in many cases. Obviously, philicity (like local softness) being a scaled Fukui function cannot provide more reliable intramolecular reactivity (when only one molecule is considered) than that is obtainable from the Fukui function but for the cases where the molecule is undergoing an intramolecular process like vibration, internal rotation, intramolecular rearrangement reaction and/or interaction with a solvent or an external field or collision with other atomlmoleculelion, etc. where both . the local and the global descri ptOfS change during · the ' physico-chemical process ,to have a different variation in philicity (like . local softness)

PARTHASARATHI el al. : QUANTUM CHEMICAL DESCRIPTORS IN CHEMOINFORMATICS 115

than that of the Fukui function even for the same site of the same molecule. The constancy of the external potential is generally tacitly assumed in all such cases. The philicity is, however, not less reliable than the Fukui function and the local softness. In addition, philicity and local softness contain information about global electrophilicity and global softness respectively. The relative electro (nucleo) philicit/O.4I, originally introduced to avoid negative values of condensed softnesses42

, is also identical with the corresponding relative Fukui function for the same molecule (and hence a poor intermolecular descriptor, devoid of any global information) even for the vario;us intramolecular physico-chemical processes described above and hence does not have any additional significance when intramolecular reactIvIty is analyzed43

. The relative electro (nucleo) philicity treats the anions and the cations at par but the anions are known to require a more sophisticated technique than the cations to produce the same quality wave functions. Also, the analysis of the radical reactions using these descriptors is not straightforward. During an electrophile-nucleophile interaction process, when two reactants approach each other from a large distance they see each other's global electrophilicities without any idea about their local counterparts. One with the larger electrophilicity will behave as an electrophile and the other as a nucleophile. The most electrophilic site of the electrophile will prefer to interact with the most nucleophilic site of the nucleophile. It may be noted that the atom with the maximum value of the local electrophilicity in the electrophile may not necessarily have larger local electrophilicity value than that of the most electrophilic atom in the nucleophile. A similar situation will arise during hard - soft interaction and will show that the local HSAB principle may not be always in conformity with its global counterpart. The Fukui function and all other related descriptors like local softness and philicity may not provide reasonable trends for the hard-hard reactions where charge based descriptors are known to be more appropriate44

. Reliability of philicity and related descriptors vis-a-vis that of relative electrophilicity has been analyzed in several papers43

-52 where the

superiority of the former has been highlighted. In the exploration of new reactivity descriptors to probe reactivity of chemical systems, the use of pronounced philicity ' based on group approach in the light of generalizedphilicity concept has been attempted.

The condensed philicity summed over a group of relevant atoms is defined as the "group philicity". It can be expressed as,

... (16)

where 'n' is the number of atoms coordinated to the

reactive atom, w: is the local electrophilicity of the

atom k, and COg a. is the group philicity obtained by adding the 10calpb!licity of the nearby bonded atoms, and (a= +, -, 0) ;'represents nucleophilic, electrophilic and radical atfticks, respectively. In the above study, we used the nucl¢ophilic group philicity index (00/) of the selected l systems to compare the chemical reactivity trends.

The electric dipole polarizability is a measure of the linear response of the electron density in the presence of an infinitesimal electric field F and it represents a second,o.rder variation in energy:

a =_(1L) _ a ,b dF;,dFj}; a,b-x,y,z. .. . (17)

~,I.

The polarizability a is calculated as the mean value as given in the following equation:

... (18)

The amount of charge transfer between any two systems can be obtained by ilPplying the formula32

. 33 :

.. . (19)

Although conceptual density functional theory has been used in numerous investigations to probe the chemical reactivity and site selectivity, their applications in the area of structure-activity relationship aspects of biochemoinformatics are limited. Some of the important contributions in this area are highlighted in the following section.

Based on the success of these OFT descriptors as revealed in the previous studies 16- 3 1 and also due to their simple calculation procedure, the usefulness of the OFT, descriptors in the QSAR and QSPR parlance has been'probed in detail.

116 INDIAN J CHEM, SEC A, JANUARY 2006

Results and Discussion Some of the bio-chemo-informatics related studies

that have l)een carried out by our group using density functional theory derived global and local quantum chemical descriptors are presented in this review article.

(a)

ellS

(b)

(c)

Fig. I---Geometries with atom numbering schemes for (a) 2,2', 5,5'-TCBP, (b) 3,3' ,4,4',5-PCBP, and (c) Benzidine.

Polychlorinated biphenyl The chemical reactivity and selectivity profiles for

22'55'-TCBP is computed and compared with those of 33'44'5-PCBp53,54. The geometries of 22'55'-TCBP and 33'44'5"PCBP (Figs 1a and Ib) were optimized by using Becke' s three parameter hybrid density functional , B3L YP/6-31 G*, which includes both Hartree-Fock exchange and DFf exchange correlation functionals55

•57

. Above calculations are carried out using the GAUSSIAN 98 package58

. The optimized geometries were characterized by harmonic vibrational frequencies which confirmed that the structure of 22'55'-TCBP is a minimum on the potential energy surface. The relative energy for 22'55'-TCBP IS calculated as

M(¢)=[E(¢)-E(¢=900)] using the total energies of

respective optimized conformations. To select proper electronic descriptor based on DFf, for the possible toxicity of the 22'55'-TCBP, the various reactivity and selectivity descriptors such as chemical hardness, chemical potential, polarizability, electrophilicity index and the local e1ectrophilic power are calculated for all the rotated conformations (Table 1). The condensed Fukui function is calculated using the natural population analysis (NPA)59. Since, Hirshfeld60 population scheme (Stockholder Partitioning Scheme) is known to provide non­negative Fukui function (FF) values, it has also been used to calculate FF values as implemented in the DMOe package61 employing BL YP/DN method. It has been found that 22'55'-TCBP has very large rotational energy barrier at <»=00 and <»=1800 with relative energy of 53.17 kllmol. Due to large rotational barrier, this molecule cannot adapt planar conformation and hence it is less toxic. In the case of 33'44'5-PCBP with very small rotational energy barrier of 7.36 kllmol at the planar orientation54 (Fig. 2a), is shown to have flexible planarity so that it

Table l-Calculated relative energy, chemical hardness, chemical potential, polarizability and electrophilicity index of2,2',5,5'-TCBP

Torsional angle Relati ve energy Chemical hardness Chemical potential Polarizabilty Electrophilicity index (degrees) (kJ/mol) (eV) (eV) (au) (eV)

-30 18.74 2.505 -4.187 168.392 3.500 0 69.52 2.405 -4.278 17l.671 3.804 30 18.92 2.505 -4.187 168.315 3.500 60 1.01 2.696 -4.078 165.014 3.085 90 0 2.911 -3 .915 163.196 2.632 120 2.30 2.709 -4.053 164.966 3.032 150 26.86 2.53l -4.166 167.808 3.428 180 122.69 2.333 -4.266 171.065 3.900 210 26.86 2.531 -4.166 167.841 3.428

PARTHASARATHI et nl.: QUANTUM CHEMICAL DESCRIPTORS IN CHEMOINFORMATICS 117

10

~> 8

~.!.

~~ 8

~~ 4

i1 O::U

0

-50

> 0.5

.!. ... C1> 0.4 ~ 0

Q,.

,g 0.3

:c Q.

~ 0.2 u G)

W C;; 0.1 u 0 oJ

0.0

-60

016 015 014 013 012

_ 011

~ 010 -= 000

'* 000 :; 007 i= 000 ~ Oel; :u 004 t5 003

000 , 001 000

0 50

It--~

0 50

(a) 1.8

1.7 ~ III III GI c "E

1.8 ~

j

1.5 ~

1 .• 100 150 200 250

(b)

---C, ---c; ---<; -.-C. -+-Cs -+-C. --H,. --H,. -H" -+-0,. -<>-0.

, 100 150 3D 250

(c) 3

6

7

4

2

5

r-~-.--~-.~~r-~~--~'-~~ o 100

Torsional Angle (degrees)

Fig. 2-{a) Variation of relative energy (kJ/mol), chemical hardness (eV) and scaled hardness (eV) with the torsional angle (deg) for 3.3',4,4'.5-PCBP. [l, scaled hardness; 2, relative energy; and. 3, chemical hardness]; (b) The variation of local electrophilic power (eV) with the torsional angle (deg) for C atoms, H atoms and CI atoms in 2,2',5.5'-TCBP using Hirshfeld [Janioning scheme, and, (c) Charge transfer between 2,2',5,5'-TCBP with various torsional angle (deg) and BaseslBase Pairs. [I, Adenine; 2, Thymine; 3, Guanine; 4, Cytosine; 5, Uracil; 6, GCWC; 7, ATH].

changes its conformation while moving in biological systems, thereby interacting readily, exhibiting its toxic properties. Moreover, the comparison between the chemical reactivity and selectivity profiles of

33'44'5-PCBP with 22'55'-TCBP including the amount of charge transfer between these toxins and biosystems modeled as nucleic acid bases and selected base pairs reveals that 33'44'5-PCBP is a highly toxic system as evident from the previous reports54. Solvation of those systems also provides the same information with only a shift in their minimum relative energy conformation (Table 2). Local electrophilic power of the indi vidual atom and possible active reactive sites were reported for 22'55'­TCBP (Fig. 2b) and compared with those of 33'44'5-PCBp54. The calculated charge transfer using Eq. (19) between the 22'55'-TCBP and NA baseslDNA base pairs shows that the charge transfer takes place from NA bases/DNA base pairs to 22'55'-TCBP (Fig. 2c). A similar calculation provided a clue that charge transfer is more in the case of 33'44'5-PCBp54 (Table 3). This calculation provides an interesting clue that 22'55'-TCBP is less toxic compared to the 33'44'5-PCBP. The electron accepting nature of PCB is evident from the charge transfer calculation.

Since the tOXICIty of halogenated aromatic hydrocarbons mainly originates from their interaction with biosystems essentially through electron transfer, it is expected that global and local electrophilicities would provide important insights into their toxic behaviour. An excellent linear correlation between the binding affinity of polychlorinated dibenzofurans with AhR receptors and a linear combination of global and local electrophilicities has been observed62

• The genesis of this beautiful quantitative structure-activity relationship is also now well understood.

Benzidine OFT based chemical reactivity descriptor analysis

has provided valuable information about the reactive sites for various types of attacks and orientations for a selected arylamine system namely, benzidine. The geometry of benzidine is depicted in Fig. Ic with the atom numbering. The geometry of benzidine is optimized by B3LYP/6-31G*, using GAUSSIAN 98 package55

.58

. The relative energy of benzidine is calculated as a function of torsional angle </J (rotation

through the C (atom No.7)-C (atom No.3) bond). To calculate the relative energy, the geometry at various </J values was optimized at the same level.

It is possible to note from the rotational energy barrier63 (Table 4), which has a small variation (0 to 11.19 kJ/mol) that this molecule is highly flexible and

118 [NDIAN J CHEM, SEC A, JANUARY 2006

Table 2--Effect of explicit solvation on the various density functional descriptors for 2i55'-TCBP using polarizable continuum model

Torsional angle Relative energy Chemical hardness Chemical potential Electrophilicity index (degrees) (kJ/mol) (eV) (eV) (eV)

-30 30.09 2.524 -4.166 3.438 0 82.99 2.422 -4.243 3.717

30 30.17 2.523 -4.096 3.325 60 11.43 2.722 -4.078 3.055 90 0.21 2.957 -3 .851 2.508 120 0.0 2.744 -4.023 2.949 150 34.89 2.559 -4.137 3.344 180 133.79 2.363 -4.269 3.857 210 34.98 2.559 -4.126 3.326

Table 3--(:aIculated charge transfer between 33'44'5-PCBP and baseslbase pairs

Torsional angle Adenine Guanine Thymine Cytosine Uracil GCWC ATH (degrees)

-30 0.107 0.132 0.037 0.076 0.023 0.139 0.102 0 0.113 0.138 0.042 0.081 0.027 0.146 0.108

30 0.107 0.132 0.037 0.076 0.023 0.139 0.102 60 0.097 0.121 0.030 0.067 0.017 0.126 0.092 90 0.091 0.114 0.027 0.062 0.014 0.117 0.086 120 0.097 0.121 0.030 0.067 0.017 0.126 0.092 150 0.107 0.132 0.037 0.076 0.023 0.139 0.102 180 0.113 0.138 0.042 0.081 0.027 0.146 0.108 210 0.107 0.132 0.037 0.076 0.023 0.139 0.102

Table 4-Calculated relative energy, chemical hardness, chemical potential, polarizability and electrophilicity index of benzidine

Torsional angle Relative energy Chemical hardness (degrees) (kJ/mol) (eV)

-30 0.0 2.33 0 6.59 2.22 30 0.0 2.33 60 4.28 2.48 90 11.19 2.61 120 4.28 2.48 150 0.0 2.33 180 6.59 2.22 210 0.0 2.33

it can adopt variety of conformations. This rota­tional freedom allows benzidine to freely interact with the cellular components in the realistic environment and hence their toxic nature. It has been found that relatively low energy barrier has provided greater flexibility to the sel~cted system, thereby allowing it to orient itself in any desired conformation in the biological system leading to its toxic characteristics. The MESP surface of benzidine63 reveals the site of attack and also provides clues for the role of electrostatic interactions involved in the reactivity. Further, the charge transfer between benzidine and nucleic acid baseslbase pairs, AHH receptors has

Chemical potential Polarizabilty Electrophilicity index (eV) (au) (eV)

-2.10 149.06 0.95 -2.13 151.97 1.02 -2.10 149.06 0.95 -2.14 143.46 0.92 -2.36 139.07 1.07 -2.14 143.47 0.92 -2.10 149.04 0.95 -2.13 152.04 1.02 -2.10 149.04 0.95

clearly revealed the electron donating nature of benzidine.

Dibenzofuran Figure 3 shows the excellent correlation between

the observed binding affinity (BA) of the toxins with AhR receptors64

•6S in terms of their [pICso= -log

(ICso)] values where ICso is the inhibitor concentration66 needed to reduce the enzyme activity by 50% and the corresponding calculated BA values

in terms of w and OJ: . The correlation is very good in

the cases of the local descriptors calculated using HPA (R=0.8742, Fig. 3a) and also by using the BA

PARTHASARATHI et al.: QUANTUM CHEMICAL DESCRIPTORS IN CHEMOINFORMATICS 119

8

7

<{ 6 CD

S c: Q)

E 5 . ~ C. x

W

4

3

2

R=O.8742 N=31

• a • • • " • • • •• •

• • •

•

• • •

• Calculated BA= -3.8314 + 613.4692·W + 15.1564·W·

3 456

Calculated SA

7

'""-'

8

9

8

~ 7 () a. S 6 c: Q)

E 'c Q) 5 C. x UJ

4

3

R=O.8722 N=31 •

b • • • " •

• • • • • •

• • •

•

• • •

• Calculated pIC50= -4.3483 + 16.632S·W + 706.SS21·W"""

3 4 5 6 8 9

Calculated plCso

Fig. 3--Correlation between calculated and experimental binding affinities (prCso): (a) HPA, (b) corresponding values normalized to that of (TCDF).

values normalized66 to that of 2, 3, 7, 8-tetrachloro dibenzofuran (TCDF) [Fig. 3b (R=O.8722, HPA)]. These correlations are better than both one and multi­parameter fits reported in reference67. Incidentally, their finding67 of reliability of softness as the toxicity descriptor is a manifestation of the maximum hardness principle27 as was highlighted in the toxicity analysis of polychlorinated biphenyls53,54 and benzidine63. The superiority of the present QSAR not only rests on better regression but also on the transparent understanding of the DFT -based global and local reactivity descriptors used here vis-a-vis the importance of the electron transfer ability between a toxin and a biosystem in gaining insights into the overall toxicity.

Group philicity

Philicity values calculated using the HPAIMPA schemes provide appropriate intermolecular reactivity trends for several sets of molecules5o. A new chemical reactivity descriptor, namely the group philicitiO, has been defined based on group a;)proach in the light of the unified philicity (Eq. 16). The global electrophilicity index contains more information about electrophilic attack when compared to the global softness and as a consequence the local Ok + has added information than the corresponding Sk + in providing corresponding intermolecular reactivity trends. The usefulness of these descriptors in explaining the reactivity trends in carbonyl compounds has been studied and the reactivity trends

have also been compared with those obtained from other local quantities including the relative electrophilicity and the group softness. Group philicity values derived from both MPA and HPA schemes have provided the expected reactivity trends in all sets of molecules considered for evaluation (Table 5). Hence, philicity and group philicity can be used as better chemical reactivity descriptors when compared to all other local reactivity descriptors.

Effect of electric field on the global and local reactivity indices of the selected aldehydes, viz. formaldehyde and acetaldehyde has been studied recently by our group68. It is interesting to note from the results that electric field significantly alters the charge distribution of the molecular systems and hence global and local reactivity trends. It is possible to highlight that the electric field does not appreciably influence the global chemical hardness whereas chemical potential is highly sensitive to the applied field. The calculated results indicate that electric field considerably affects all the local reactivity indices and hence the site of attack and selectivity.

A systematic investigation69 has also been made to study the effect of solvation on the local philicity indices using B3L YP scheme employing direct calculation method70. Due to the negativity of the FF indices calculated using direct approach, the results have also been compared with those values obtained from Hirshfeld population scheme. It is possible to observe from the results that solvation marginally influences the local reactivity profiles69. We have also

120 INDIAN J CHEM, SEC A. JANUARY 2006

Table 5 - Calculated local molecular reactivity properties of the selected compounds and on carbonyl carbon of the same in MPA/HPA scheme

Systems

HCHO CH3CHO CH3COCH3

C2HsCOC2Hs

C6HsCHO p-MeOC6H4CHO

CH2=CHCHO CH3CH=CHCHO C6Hs=CHCHO

CH)COCI CH3COOCH3

0.303 0.287 0.246 0.251

0.156 0.141

0.152 0.155 0.096

0.033 0.303

Sk+ Group Sk+1 Sk' ~+

softness Mulliken Population Analysis

0.019 0.018 0.016 0.016

0.013 0.012

0.012 0.012 0.009

0.002 0.017

0.063 0.040 0.021 0.019

0.040 0.039

0.038 0.038 0.D21

0.046 0.028

0.577 0.590 0.557 0.563

0.366 0.345

0.358 0.367 0.248

0.164 0.546

0.298 0.222 0.165 0.157

0. 168 0.115

0.186 0.161 0.107

0.037 0.194

(J)+ g

0.983 0.493 0.221 0.186

0.501 0.366

0.615 0.508 0.243

0.823 0.316

0.397 0.300 0.211 0.135

0.142 0.142

0.206 0.174 0.108

0.233 0.129

Sk+ Group S k+1 Sk' ~+

softness Hirshfeld Population Scheme

0.102 0.072 0.049 0.031

0.043 0.042

0.067 0.055 0.039

0.047 0.023

0.258 0.185 0.137 0.089

0.127 0.119

0.158 0.170 0.132

0.159 0.072

0.170 0.1 29 0. 109 0.090

0.080 0.070

0.154 0.111 0.083

0.076 0.079

1.856 1.027 0.590 0.355

0.783 0.642

1.300 0.933 0.733

0.952 0.258

(J)+ g

4.673 2.654 1.630 1.010

2.303 1.847

3.075 2.902 2.490

3.246 0.818

3.5,----------------------------------------------------------. 3

~ 2.5 :Q 2 :c a. 1.5

~ 1 o ....J 0.5

(b) (c)

f---.. r; (G)

---. .. r;(G)

-t- .. r:(G)

~ .. r:(s) -.-.. r; (S)

-+- .. ':(S)

o -0.5

C I 02 H3"4 C 1 C 2 H 3 H" H 5 0 6 H 7 C 1 0 2 C 3 C" H 5 H 6 H 7 H 8 H 9 H 10 C 1 C 2 H 3 H" C!5 H 6 0 7 H 8

Atom

Fig. 4--Local philicity of: (a) Formaldehyde, (b) Acetaldehyde, (c) Acetone, (d) Acrolein using direct calculation [(G)-Gas phase, (S)-Solution phase].

made an attempt to verify the unified philicity indices proposed by Chattaraj et al. It is evident from the results that (f)u (where a= +, -, and 0) values could describe effectively the reactivity as well as site selectivity when compared to the other local quantities in gas phase as well as in solvent environment (Fig. 4). The solvent continuum around the solute molecules marginally influences the local quantities. Maximum changes in the local quantities are observed in acetone whereas minimum variations are noted in formaldehyde. The FF indices derived from BL YPIDNO ITirshfeld population scheme using conventional FF method are positive and marginal increase in the local quantities are also observed from the calculations based on solvent environment.

QSAR I QSPR studies

The biological activity of testosterone and estrogen derivatives has been analyzed by our group using

electrophilicity index as a descriptor7 t• It is known

from the earlier studies3.

tO on QSAR that it is neces~ary to use various possible combinations of structural descriptors. In this context, the SAR based on electrophilicity has been shown to be promising. Since the electrophilicity index is a chemical reactivity descriptor and its definition has strong foundation from the density functional theory, it is appropriate to make use of this descriptor in the QSAR parlance and the usefulness of such application was evident from our investigation. Results emanated from that study (Tables 6 and 7) showed that the electrophilicity can be used as a descriptor of biological activity and it is unbelievable that a single descriptor can provide such a beautiful correlation.

In order to test the efficacy of this new concept, OF]' level calculations with B3L YP exchange­correlation functional and 6- 31 +G* basis set using GAUSSIAN 03 program are performed43 on several

PARTHASARATHI et al.: QUANTUM CHEMICAL DESCRIPTORS IN CHEMOINFORMATICS 121

Tab.le 6-Electrophilicity index of testosterone derivatives with their observed and calculated biological activity in terms of relative binding aftinity (RBA)

No. Set-l Electrophilicity Electrophilicity Expt.a Calc. RBA (AMI) (B3LYP) RBA AMI B3LYP

I 19-Nortestosterone 0.092 0.099 0.40 0.689 0.735 2 F1uoxymesterone 0.090 0.096 0.77 0.754 0.819 3 17a-methyltestosterone 0.091 0.100 0.85 0.701 0.732

4 Oxymetholone 0.076 0.080 1.54 1.314 1.159 5 Ethylestrenol 0.055 0.037 2.00 2.102 2.114

'Experimental data as given in Ref. 78

Table 7--Electrophilicity index of 16a-substituted estradiol derivatives with thdr observed and calculated biological activity

No Set-I Electrophilicity Electrophilicity Expl'. Calc. RBA (AMI) (B3LYP) RBA AMI B3LYP

I H 0.070 0.050 2 1.696 1.989 2 CHzBr 0.070 0.050 1.97 1.539 1.829 3 OH 0.071 0.051 1.28 0.899 1.515 4 CHzCH=CHCH2OC6Hs 0.070 0.054 0.85 1.708 0.420 5 CH2CH2CH2CN 0.072 0.054 -0.05 0.206 0.294

' Experimental data as given in Ref. 7C)

Table 8--Proton affinity, electrophilicity, Fukui function, WN- and tiJc+

Amine I'roton affinity Electrophilicity f;; (a.u)" f ;; (a.u)b f ;; (a.u)C WN - (a.u) + We

(kcaJ/mol) ill (a.u) (a.u)

NH3 207.0 0.055369 0.9764 0.9743 0.0539 CH3NHz 218.4 0.058047 0.3173 0.8135 0.8276 0.0480 0.1200 CH3CH2NH2 221.4 0.055959 0.3105 0.7731 0.8256 0.0462 0.1522 (CH3hNH 224.8 0.051572 0.3017 0.7401 0.7341 0.0378 0.0532 (CH3)3N 228.6 0.049094 0.2902 0.6891 0.6506 0.0319 0.0259 (CH3CH2hNH 229.4 0.055257 0.7187 0.0397 0.0949 (CH3CH2)3N 235.5 0.046986 0.6316 0.0296 0.0832 (CH3hN(CH3CH2) 230.9 0.047796 0.6427 0.0307 0.0355 (CH3)NH(CH3CH2) 227.3 0.049196 0.7257 0.0357 0.0706 (CH3)N(CH3CH2h 233.3 0.047074 0.6355 0.0299 0.0639 (CH3CHzCH2)NH2 222.8 0.0.'56893 0.8048 0.0457 0.1385 (CH)CH2CH2hNH 231.4 0.049294 0.7056 0.0347 0.0787

a Data as in Ref erence 35; b Data as in Ref erence; 70 C Present work

amines . Necessary Fukui functions ale calculated using the direct method proposed by Contreras et al.70

. The molecules considered are NH3, CH3NH2,

CH3CH2NH2, (CH3CH2CH2) NH2, (CI-hCH2)2NH, (CH3hNH, (CH3)NH(CH3CH2), (CH3CI-hCH2hNH, (CH3)3N, (CH3hN(CH3CH2), (CH3)N(CH3CH2h , (CH3CH2)3N, NH20H, CH3NHOH, CI-hCI-hNHOH, (CH3hNOH, (CH3CH2)2NOH, CH3CH2NOHCH3 and CH3CH2CH2NOH. For all the molecules GJk- is the highest in the N-center indicating that this ~:ite would be most favourable for electrophilic attack (say protonation). As shown in Table 8 the order in which

())N varies is NH3 > CH3NH2 > CH3CH2NH2 > (CI .. hCI-lzCH2) NH2 >(CH3CH2)2NH > (CH3hNH > (CH])NH(CH3CH2) > (CH3CH2CH2hNH > (CH3hN > (CI'h)2N'(CH3CH2»(CH3)N(CH3CH2h > (CH3CH2)3N which corroborates with other such calcula-

. 357072-76 h ff" f I' h . hans .. on gas p ase proton a mlty a a lp atrc amines .

In Table 8, the numerical data of gas phase proton

affinity, electrophilicity, f;;, 04;- and ())c+ values are

given . Our f;; values are slightly different from that

of Contreras et al. 70, because we have used B3L YP/6-

122 INDIAN J CHEM, SEC A, JANUARY 2006

235 R=O.96707. N=12

• • • 230

• • <t 225 0.. (ij • U ~ 220 • • 0 Q) .c ~ 215

210

205 205 210 215 220 225 230 235 240

Experimental PA

Fig. 5-Correlation between the observed proton affinity (kcal/mol) of amines in Table 8 and those calculated from a two­parameter linear model (Eq. 20).

Table 9--The W, WN- and CLb- of hydroxylamine derivatives

Molecule w(a.u) WN- (a.u) wo-(a.u)

NH20H 0.0703 0.0574 0.0075 CH3NHOH 0.0680 0.0477 0.0067 CH)CH1NHOH 0.0648 0.0445 0.0066 (CH3)2NOH 0.0607 0.0389 0.0034 (CH3CH2hNOH 0.0596 0.0381 0.0034 CH3CH1NOHCH) 0.0595 0.0377 0.0034 CH3CH2CH2NOH 0.0651 0.0443 0.0069

31 +G* basis set whereas they have used HP/6-311 G basis set. An inverse relationship has been observed between gas phase basicit/4 and WN- (Refs 35, 70, 72, 73). Proton affinity can be linearly correlated (R = 0.9671) with one global (w) and one local (WN-)

variables as shown in Fig. 5. The corresponding equation is given by,

PACalculatcd =218.R292-1473.5770*w; + 1234.7735*w

.. . (20)

which is in the similar spirit as that of Yang and Mortier35

. Corresponding experimental proton affinities are taken from Ref. 75.

On the other hand, w/ is the highest in the C­centers signifying that the nucleophilic attack will take place in the C-site. The order of preference turns out to be CH3CH2NH2> (CH3CH2CH2)NH2 > CH3NH2 > (CH3CH2hNH > (CH3CH2)3N > (CH3CH2CH2)2NH > (CH3)NH(CH3CH2) > (CH3)N(Cl-hCH2h > (CH3hNH > (CH3hN(CH3CH2)

> (CH3hN. For hydroxylamine and its derivatives WN-

Table 1 O--Abbreviations of the selected alkanes and their isomers (cz-cs)

Abbreviation

2 3 4 2M3 5 2M4 22MM3 6 2M5 3M5 23MM4 22MM4 7 2M6 3M6 3E5 24MM5 22MM5 23MM5 33MM5 223MMM4 8 1M7 3M7 4M7 25MM6 3E6 24MM6 22MM6 23MM6 34MM6 33MM6 2M3E5 224MMM5 234MMM5 3M3E5 223MMM5 233MMM5 2233MMMM4

rUPAC name

Ethane Propane Butane

2-methyl propane Pentane

2-methyl butane 2,2-dimethyl propane

Hexane 2-methyl pentane 3-methyl pentane

2,3-dimethyl butane 2.2-dimethyl butane

Heptane 2-methyl hexane 3-methyl hexane

3-ethyl pentane 2,4-dimethyl pentane 2,2-dimethyl pentane 2,3-dimethyl pentane 3,3-dimethyl pentane

2,2,3-trimethyl butane Octane

2-methyl heptane 3-methyl heptane 4-methyl heptane

2,5-dimethyl hexane 3-ethyl hexane

2A-dimethyl hexane 2,2-dimethyl hexane 2,3-dimcthyl hexane 3,4-dimethyl hexane 3,3-dimethyl hexane

2-methyl-3-ethyl pentane 2,2,4-trimethyl pentane 2,3,4-trimethyl pentane

3-methyl-3-ethyl pentane 2,2,3-trimethyl pentane 2.3,3-trimethyl pentane

2,2,3,3-tetramethyl butane

is the largest and Wo- is the second largest (Table 9) which is in conformity with the experimental results 76,

that state that the N-center is the most suitable for protonation.

Ionization potential (IP) can also be used as a descriptor to understand structure-activity and structure-property relationship. Within the framework of Hartree-Fock theory, the computed IP has excellent correlation with the macroscopic properties77 such as boiling point (BP), heat of formation or enthalpy, entropy, heat capacity and heat of vapourization (Tables 10 and 11, see Ref. 77 for details). The correlation coefficient has been found to be high for

PARTHASARATHI et al.: QUANTUM CHEMICAL DESCRIPTORS TN CHEMOINFORMATICS 123

Table 11---Experimental" and calculated values of five selected macroscopic properties of C2-CS alkanes and their isomers using ionization potential as descriptorb

Alkanes I (eV)

!:J.Hro

(Kcal / mol) S'

(Cal/ K mol.) Cp

(Cal/ K mol) at 25°C

!:J.H v

(Cal/mol)

Exp. Cal. Exp. Cal. Exp. Cal. Exp. Cal. Exp. Cal.

2 3 4 2M3 '5 2M4 22MM3 6 2M5 3M5 23MM4 22MM4 7 2M6 3M6 3E5 24MM5 22MM5 23MM5 33MM5 223MMM4 8 2M7 3M7 4M7 25MM6 3E6 24MM6 22MM6 23MM6 34MM6 33MM6 2M3E5 224MMM5 234MMM5 3M3E5 223MMM5 233MMM5 2233MMMM4

0.48623 0.46888 0.45585 0.45858 0.44584 0.44527 0.45101 0.43734 0.43817 0.43666 0.43669 0.43723 0.43102 0.43187 0.42981 0.43068 0.43032 0.43182 0.42885 0.42977 0.42916 0.42612 0.42679 0.42532 0.42432 0.42803 0.42555 0.42361 0.42678 0.42362 0.42183 0.42401 0.42587 0.42449 0.42392 0.42292 0.4236 0.42357 0.42325

-88.630 -42.070 -0.500

-11.730 36.074 27.852 9.503

68.740 60.271 63.282 57.988 49.741 98.427 90.052 91.850 93.475 80.500 79.197 89.784 86.064 80.882 125.655 117.647 118.925 117.709 109.103 118.534 109.429 106.840 115.607 117.725 111.969 115.650 99.238 113.467 118.259 109.840 114.760 106.470

-103.8 -43.6

1.7 -7.8 36.5 38.4 18.5 66.0 63.1 68.4 68.3 66.4 88.0 85.0 92.2 89.2 90.4 85.2 95.5 92.3 94.5 105.0 102.7 107.8 111.3 98.4 107.0 113.8 102.7 113.7 119.9 112.4 105.9 110.7 112.7 116.2 113.8 113.9 115.0

-20.24 -24.82 -29.81 -32.10 -35.00 -36.50 -40.20 -39.96 -41.50 -40.90 -42.10 -43.90 -44.89 -46.40 -45 .90 -45.20 -48 .10 -49.00 -46.30 -47.60 -48.60 -49.82 -51.50 -50.82 -50.69 -53 .21 -50.40 -52.44 -53.71 -51.13 -50.91 -52.61 -50.48 -53.57 -51.97 -51.38 -52.61 -51.73 -53.99

a Experimental data as given in Ref. 77. ~: methyl; E: ethyl.

the relationship between IP and BP. Hardness and softness indices exhibit similar correlation coefficient in the range of 0.80-0.95 for all the macroscopic properties, which confirms the fact that both the microscopic properties are interrelated. Maximum correlation coefficient for both the indices has been found to be 0.945 for heat of formation. This observation reinforces the existing fact that, hardness (11) holds direct relationship with the stability of a molecule. Other descriptors do not show promising

-15.8 -25 .7 -33.1 -31.5 -38.8 -39.1 -35.8 -43.6 -43.1 -43.9 -43.9 -43.7 -47.2 -46.7 -47.9 -47.4 -47.6 -46.8 -48.4 -47.9 -48.3 -49.9 -49.6 -50.5 -51.0 -48.9 -50.3 -51.4 -49.6 -51.4 -52.4 -51.2 -50.1 -50.9 -51.2 -51.8 -51.4 -51.4 -51.6

54.85 64.51 74.10 70.42 83.27 82.10 73.10 92.45 91.00 91.50 87.40 85.60 101.64 100.40 101.80 98.30 94.80 93.80 99.00 95.10 91.60 110.82 108.81 110.32 108.35 104.93 109.51 106.51 103.06 106.11 107.15 104.70 105.43 101.62 102.99 103.48 101.62 103.14 94.34

47.7 63.4 75.3 72.8 84.4 84.9 79.7 92.1 91.3 92.7 92.7 92.2 97.8 97 .1 98.9 98.1 98.5 97.1 99.8 98.9 99.5 102.3 101.7 103.0 103.9 100.5 102.8 104.6 101.7 104.6 106.2 104.2 102.5 103.8 104.3 105.2 104.8 104.6 104.9

12.5 17.4 23.3 23.1 28.6 28.4 28.8 34.0 33.9 33.4 33.3 33.8 39.4 39.3 39.1 39.6 40.8 39.8 38.4 39.6 39.0 44.8 44.7 44.4 44.7

45.0

45.6

45.0

7.6 17.6 25 .1 23.5 30.8 31.1 27.8 35.7 35.2 36.1 36.1 35.8 39.3 38.8 40.0 39.5 39.7 38.9 40.6 40.0 40.4 42.1 41.8 42.6 43.2 41.0 42.5 43 .6 41.8 43 .6 44.6 43.3 42.3 43. 1 43.4 43.9 43.6 43.6 43.8

3739.5 4811.8 5801.2 5416.2 6595.1 6470.8 5648.6 7627.2 7676.6 7743.9 7120.0 7271.0 8409.6 8538.7 8596.3 8642.8 8167.1 8106.7 8390.9 8145.4 7767.1 9221.0 9362.0 9432.0 9404.8 9110.2 9416.3 9086.6 8927.8 9224.9 9239.4 9065 .2 9134.3 8548.0 8988.2 9028.7 8861.1 8960.9 10351.5

2867.0 4596.9 5896. 1 5623.9 6894.2 6951.1 6378.7 7741.7 7659.0 7809.5 7806.5 7752.7 8371.9 8287.1 8492.5 8405.8 8441.7 8292.1 8588.3 8496.5 8557.3 8860.5 8793.7 8940.2 9039.9 8670.0 8917.3 9110.7 8794.6 9109.7 9288.2 9070.8 8885.4 9022.9 9079.8 9179.5 9111.7 9114.7 9146.6

relation with the selected macroscopic properties. The present investigations show that ionization

potential can be used as a successful quantum chemical descriptor for developing QSARlQSPR models while hardness index shows some promising results for specific macroscopic properties.

Conclusions The success of DFf based global and local

quantum chemical descriptors III predicting the

124 INDIAN J CHEM, SEC A, JANUARY 2006

chemical reactivity and selectivity profiles of several systems selected by our group are highlighted in this review. The simple calculation procedure and the usefulness of all Off based descriptors in the QSAR and QSPR parlance have also been probed in detail. In this study, the applications of global and local descriptors in the development of QSAR and QSPR have been presented for prediction of physical properties of series of alkanes, biological activity of testosterone and estrogen derivatives and toxicity of polychlorinated biphenyls, polychlorinated dibenzofurans and benzidines. It is shown that the global descriptors such as electrophilicity and ionization potential are capable of predicting the biological activity of the selected molecules and local descriptors such as philicity and group philicity are capable of identifying the activity of a particular site in the molecule and in analyzing its biological activity /toxicity as well as its behaviour during an intermolecular reaction.

Acknowledgements We are thankful to CSIR, New Delhi for financial

assistance and Drs P Thanikaivelan, J R Rao, B U Nair, T Ramasami and B Maiti for collaboration.

References 1 Altman R B & Dugan J M, Meths Biochem Annal. 44 (2003)

3. 2 Hanseh C, Hoekman D, Leo A. Weininger D & Selassie C,

Chem Rev. 102 (2002) 783. 3 Hansch C, Kurup A, Garg R & Gao H, Chelll Rev, 10 I

(2001) 619. 4 Hansch C, Maloney P P, Fujita T & Muir R M, Nature, 194

(1962) 178. 5 Hansch C, Li R I, Blaney J M & Langridge K, J Med Cllelll,

25 (1982) 777. 6 Leach A R & Gillet V J, AI! Introduction to

Chemoinformatics (Kluwer, Dordrecht) 2003. 7 Gasteiger J & Engel T, Chemoinformatics. A Textbook

(Wiley- VCH, Weinheim), 2003 . 8 Ormerod A, Willett P & Bawden D. Quant Slmct Activ

Relat, 8 (1989) 115. 9 Edgar S J, Holliday J D & Willett P, J Mol Graph Model. 18

(2000) 343. 10 Willett P, Barnard J M & Downs G M. J Chem Inf Compllt

Sci, 38 (1998) 983. 11 Kobayashi S, Hamashima H, Kurihara M. Miyata N &

Tanaka A, Chem Pharm Bull, 46 (1998) 1108. 12 MacDonal P Pd J C, Ciolino H P & Yeh G C, Cal1~e,. Res, 64

(2004) 429 . 13 Matsumoto M, Imagawa M & Masayoshi A Y, Biochem J.

338 (1999) 599; Kremer J, Gleiehmann E & Esser C. J lmmu, 153 (1994) 2778.

14 Maynard A T & Covell D G, JAm Chem Soc, 123 (200 1) 1047.

15 Parr R G, Szentpaly Lv & Liu S. J Alii Chem Soc. 121 (\ 999) 1922.

J 6 Chattaraj P K, Nath S & Maiti B, Reactivity Descriptors in Computalional Medicinal Chemistry for Drug Discovery, edited by J Tollenaere, P Bultinck, H D Winter & W Langenaeker (Marcel Dekker: New York) 2003, Chap. II, pp. 295-322. .

17 Geerlings P, De Proft F & Langenaeker W, Chem Rev. 103 (2003) 1793.

18 Balawender R, Safi B & Geerlings P, J Phys Chem A, 105 (200 I) 6703 .

19 Vivas-Reyes R, De Proft F, Biesemans M, Willem R & Geerlings P, J Phys Chem A, 106 (2002) 2753.

20 Roos G, Messens J, Loverix S , Wyns L & Geerlings P. J Phys Chem B, 108 (2004) 17216.

21 De Proft F & Geerlings P, Chem Rev, 101 (2001) 1451. 22 Chattaraj P K & Sengupta S, J Phys Chem, 100 (1996) 16126. 23 Chattaraj P K & Sengupta S, J Phys Chem A, 101 (1997)

7893; Chattaraj P K & Sengupta S, J Phys Chelll A, 103 (1999) 6122.

24 Chattaraj P K & Poddar A, J Phys Chem A, 102 (1998) 9944; Chattaraj P K & Poddar A, J Phys Chem A, 103 (1999) 1274; Chattaraj P K & Poddar A, J Phys Chem A, 103 (1999) 8691.

25 Chattaraj P K & Maiti B, J Phys Chem A, 105 (2001) 169; Chattaraj P K & Maiti B, J Phy.I' Chem A, 108 (2004) 658; Chattaraj P K & Maiti B, JAm Chem Soc, 125 (2003) 2705; Chattaraj P K & Maiti B, J Chem Educ, 78 (2001) 811.

26 Chattaraj P K & Sarkar U, J Phys Chen! A, 107 (2003) 4877. 27 Chanar ... j P K & Schleyer Pv R, J Alii Chell! Soc, 116 (1994)

1067; Parr R G & Chattaraj P K, JAm Chell! Soc, 113 (1991) 1854; Chattaraj P K, Lee H & Parr R G, JAm Chem Soc, 113 (1991) J 855; Chattaraj P K, Fuentealba P, Gomez B & Contreras R, JAm Chem Soc, 122 (2000) 348; Chattaraj P K, Fuentealba P, Jaque P & Toro-Labbe A, J Phys Chelll A, 103 (1999) 9307; Chattaraj P K, Cedillo A, Parr R G & Arnett E M, J Org Chem, 60 (1995) 4707; Chattaraj P K, Gutierrez­Oliva S, Jaque P & Toro-Labbe A, Mol Phys, 101 (2003) 2841.

28 Chattaraj P K, Nath S & Sannigrahi A B, Chem Phys Lett, 212 (1993) 223; Chattaraj P K, Nath S & Sannigrahi A B. J Phys C!1eI/!, 98 (1994) 9143.

29 Chattaraj P K, Maiti B & Sarkar U, J Phys Chem A, 107 (2003) 4973.

30 Thanikaivelan P, Padmanabhan J, Subramanian \f & R:unasami T, Theor Chem Ace. 107 (2002) 326; Parthasarathi R, Madhan B, Subramanian V & Ramasami T, Th eol' Chem Acc, 110 (2003) 19; Subramanian V, Sivanesan D, Amutha R, Padmanabhan J & Ramasami T, Chem Phys Lett, 294 (1998) 285; Sivanesan D, Subramanian V & Nair B B, J MoL. StrLlct, 544 (2001) 123.

31 Arulmozhiraja S, Fujii T & Salo G, Mol Phys, 100 (2002) 423 ; Arulmozhiraja S, Selvin PC & Fujii T , J Phys Chen! A, 106 (21)02) 1765.

32 Pan- R G & Yang W, Density Functio/lal Theory of Atoms ((/ld Molecules (Oxford University Press, Oxford) 1989.

33 Pearson R G, Chemical Hardness--Applications from Molecules 10 Soiids, (VCH-Wiley: Weinheim) 1997; Parr R G & Pearson R G, JAm Chem Soc, 105 (1983) 7512.

34 Parr R G & Yang W, JAm Chem Soc, 106 (1984) 4049; Fukui K, Science, 218 (1987) 747; Ayers P W & Levy M, Th ea Chem Acc, 103 (2000) 353.

PARTHASARATHI et af.: QUANTUM CHEMICAL DESCRIPTORS IN CHEMOINFORMATICS 125

35 Yang W & Mortier W J, J Arn Chern Soc, 108 (1986) 5708. 36 Lee C, Yang W & Parr R G, J Mol Struct (Theochern), 163

(1988) 305. 37 Cioslowski J, Martinov M & Mixon S T, J Phys Chern, 97

(1993) 10948. 38 Yang W & Parr R G, Proc Natl Acad Sci, 82 (1985) 6723. 39 Krishnamurti S & Pal S, J Phys Chem A, 104 (2000) 7639. 40 Roy R K, J Phys Chern A, 108 (2004) 4934. 41 Roy R K, Krishnamurti S, Geerlings P & Pal S, J Phys Chem

A, 102 (1998) 3746. 42 Geerlings P, De Proft F & Langenaeker W, Chem Rev, 103

(2003) 1812. 43 Chattaraj P K & Sarkar U, Proc of IECMD 2003. 44 Chattaraj P K, J Phys Chem A, 105 (2001) 511; Melin J,

Aparicio F, Subramanian Y, Galvan M & Chattaraj P K, J Phys Chern A, 108 (2004) 2487; Hocquet A, Toro- Labbe' A & Chermette H, J Mol Struct (Theochern), 686 (2004) 213.

45 Domingo L R, Aurell M J, Pe'rez P & Contreras R, J Phys Chern A, 106 (2002) 6871; Pe'rez P, Toro-Labbe' A, Aizman A & Contreras R, J Org Chern, 67 (2002) 4747; Domingo L R, Aurell M J, Pe'rez P & Contreras R, J Org Chern, 68 (2003) 3884.

46 Chamorro E, Chattaraj P K & Fuentealba P, J Phys Chern A, 107 (2003) 7068.

47 Mananes A, Duque F, Mendez F, Lopez M J & Alonso J A, J Chem Phys, 119 (2003) 5128 (In this paper it is shown that the Hirshfeld Population Analysis provides negative Fukui functions in some cases).

48 Olan J, van Alsenoy C & Sannigrahi A B, J Phys Chern A, 106 (2002) 3885. Tishchenko 0, Pham-Tran N, Kryachko E S & Nguyen T M, J Phys Chern A, 105 (2001) 8709. Fuentealba P & Contreras R in Reviews of Modem Quantum Chernistry; edited by K 0 Sen (World Scientific, Singapore) 2002, pp 10 13-1052.

49 Meneses L, Tiznado W, Contreras R & Fuentealba P, Chern Phys Lett, 383 (2004) 181; See also Domingo L R, Pe'rez P & Contreras R, Tetrahedron, 60 (2004) 6585, where it has been argued that the local electrophilicity can elucidate the intramolecular selectivity properly.

50 Parthasarathi R, Padmanabhan J, Elango M, Subramanian Y & Chattaraj P K, Chem Phys Lett, 394 (2004) 225.

51 Morrell C, Grand A & Toro-Labbe A, J Phys Chem A, 109 (2005) 205.

52 Chatterjee A, Ebina T & Iwasaki T, J Phys Chern A, 105 (2001) 10694.

53 Parthasarathi R, Padmanabhan J, Subramanian Y, Maiti B & Chattaraj P K, J Phys Chern A, 107 (2003) 10346.

54 Parthasarathi R, Padmanabhan J, Subramanian Y, Maiti B & Chattaraj P K, Curr Sci, 86 (2004) 535.

55 Becke A 0, Phys Rev A, 38 (1998) 3098. 56 Lee C, Yang W & Parr R G, Phys Rev B, 37 (1988) 785;

Miechlich B, Savin A, Stoll H & Preuss H, Chem Phys Lett, 157 (1989) 200.

57 Hariharan P C & Pople J A, Theor Chirn Acta, 28 (1973) 213.

58 Frisch M J, Trucks G W, Schlegel H B, Scuseria G E, Robb M A, Cheeseman J R, Zakrzewski Y G, Montgomery J A Jr, Stratmann R E, Burant J C, Dapprich S, Millam J M, Daniels

AD, Kudin K N, Strain M C, Farkas 0, Tomasi J, Barone Y, Cossi M, Cammi R, Mennucci B, Pomelli C, Adamo C, Clifford S, Ochterski J, Peters son G A, Ayala P Y, Cui Q, Morokuma K, Malick OK, Rabuck A 0, Raghavachari K, Foresman J B, Cioslowski J, Ortiz J Y, Baboul A G, Stefanov B B, Liu G, Liashenko A, Piskorz P, Komaromi I, Gomperts R, Martin R L, Fox 0 J, Keith T, AI-Laham M A, Peng C Y, Nanayakkara A, Gonzalez C, Challacombe M, Gill P M W, Johnson B, Chen W, Wong M W, Andres J L, Gonzalez C, Head-Gordon M, Replogle E S & Pople J A, Gaussian 98, Revision A.7, (Gaussian, Inc. ; Pittsburgh, USA) 1998.

59 Reed A E, Curtiss L A & Weinhold F, Chem Rev, 88 (1998) 899.

60 Hirshfeld F L, Theor Chirn Acta, 44 (1977) 129.

61 DMOL3, Accelrys. Inc.; San Diego, California, USA.

62 Sarkar U, Parthasarathi R, Subramanian Y & Chattaraj P K, (Submitted for publication).

63 Parthasarathi R, Padmanabhan J, Subramanian Y, Sarkar U, Maiti B & Chattaraj P K, Int Eletron. J Mol Des, 2 (2003) 798.

64 Bandiera S, Sawyer T W, Campbell M A, Fujita T & Safe S, Biochem Pharrnacol, 32 (1983) 3803.

65 Bandiera S, Sawyer T W, Romkes M, Zmudzka B, Safe L, Mason G, Keys B & Safe S, Toxicology, 32 (1984) 131.

66 Waller C L & McKinney J 0, Chem Res Toxicol, 8 (1995) 847.

67 Arulmozhiraja S, Morita M, Chern Res Toxicol, 17 (2004) 348.

68 Parthasarathi R, Subramanian Y & Chattaraj P K, Chern Phys Lett, 382 (2003) 48.

69 Padmanabhan J, Parthasarathi R, Sarkar U, Subramanian Y & Chattaraj P K, Chern Phys Lett, 383 (2004) 122.

70 Contreras R R, Fuentealba P, Galvan M & Perez P, Chem Phys Lett, 304 (1999) 405.

71 Parthasarathi R, Subramanian Y, Roy 0 R & Chattaraj P K, Bioorg Med Chem, 12 (2004) 5533.

72 Safi B, Choho K, De Proft F & Geerlings P, Chell! Phys Lett, 300 (1999) 85.

73 Perez P, Simon-Manso Y, Aizman A, Fuentealba P & Contreras R, J Arn Chern Soc, 122 (2000) 4756.

74 Lias S G, Liebman J F & Levin R 0, J Phys Chern Ref Data, 13 (1984) 695.

75 Aue 0 H, Webb H M & Bowers M T, J Am Chern Soc, 98 (1976) 311.

76 Angelli F, Aschi M, Cacace F, Pepi F & de Petris G, J Phys Chern, 99 (1995) 6551; Cotton F A & Wilkinson W, Advanced Inorganic Chemistry (Wiley, New York) 1980, p.22.

77 Thanikaivelan P, Subramanian Y, Rao J R & Nair B U, Chern Phys Lett, 323 (2000) 59.

78 Saartok T, Dahlberg E & Gustafsson J A, Endocrinology, 114 (1984) 2100.

79 Gao H, Katzenellenbogen J A, Garg R & Hansch C. Chern Rev, 99 (1999) 723.