# QSAR property relationship

When the property being described is a physical property, such as the boiling point, this is referred to as a quantitative structure-property relationship (QSPR). When the property being described is a type of biological activity, such as drug activity, this is referred to as a quantitative structure-activity relationship (QSAR). Our discussion will first address QSPR. All the points covered in the QSPR section are also applicable to QSAR, which is discussed next. [c.243]

A wide field of applications for chemical data mining is drug design. In short, drug design starts with a compound which has an interesting biological profile and optimizes the compound as well as its activity (see Section 10.4). Thus, the information about the biological activity of a compound is a crucial aspect in drug design. The relationship between a structure and its biological activity is represented by so-called quantitative structure-activity relationships (QSAR) (see Section 10.4). The field of QSAR can be approached via chemical data mining Starting from the structure input, e.g., in the form of a connection table (see Section 2.5), a 2D or 3D model of the structure is calculated. Ensuing secondary information, e.g., in the form of physicochemical properties such as charges, is generated for these structures. The enhanced structure model is then the basis for calculating a descriptor, i.e., a structure code in the form of a vector to which computational methods, for example statistical methods or neural networks, can be applied. These methods can then fulfill various data mining tasks such as classification or establishment of QSAR models which can finally be employed for the prediction of properties such as biological activities. [c.474]

Quantitative Structure—Activity Relationship Design. Increasing economic pressures toward more, better, and cheaper pesticides have led to the development and appHcation of the Quantitative Stmcture—Activity Relationship (QSAR) paradigm and related experimental design principles for pesticides (4). Theoretically, quantitative deterrnination of the relationships between chemical stmcture and biological and environmental properties of a molecule should permit the design of a novel molecule with exacdy those properties considered ideal for the intended appHcation. In 1964, using earlier studies of the relationships of chemical composition to biological activity (5), the use of multivariate linear regression analyses to develop the quantitative stmcture—activity rules requited for successful QSAR appHcation was outlined (6). Subsequently, QSAR has been refined, expanded, and successfully appHed in the chemorational design of pesticidal compounds (3,7—11). [c.39]

Quantitative Structure—Activity Relationships. Many quantitative stmcture—activity relationship (QSAR) studies of progestins have appeared in the Hterature and an extensive review of this work is available (174). QSAR studies attempt to correlate electronic, steric, and/or hydrophobic properties to progestational activity or receptor binding affinity. A review focusing on the problems associated with QSAR of steroids has been pubUshed (175). [c.220]

But why are 3D structures needed. Part of the answer to this question has already been given. As mentioned previously, a large variety of physical, chemical, and biological properties of a molecule are strongly dependent on its 3D structure. Therefore, studies which try to correlate chemical structures with a certain property under consideration - so-called QSAR/QSPl studies (Quantitative/Quahtative Structure-Activity/Property Relationship) - may gain more insight into the problem under investigation if 3D structural information is used. Modeling and prediction of biological activity, virtual screening and docking experiments (prediction of receptor/ligand interactions and complexes in biological systems), or investigations to model the chemical reactivity of a compound clearly require information on the 3D structure of the molecules imder consideration. In addition, the results of structure elucidation techniques which are based on experimental data, such as those obtained from X-ray crystallography, NMR or IR spectra, depend heavily on the quality of the initial geometries of the molecules during the structure refinement procedure. Furthermore, quantum mechanical or molecular mechanical calculations need at least a crude 3D molecular model as starting geometry. [c.96]

A quantitative structure-activity relationship (QSAR) relates numerical properties of tl molecular structure to its activity by a mathematical model. The term quantitative stru ture-property relationship (QSPR) is also used, particularly when some property oth( than biological activity is concerned. In drug design, QSAR methods have often bee used to consider qualities beyond in vitro potency. The most potent enzyme inhibitor is ( little use as a drug if it cannot reach its target. The in vivo activity of a molecule is often composite of many factors. A structure-activity study can help to decide which featurt of a molecule give rise to its overall activity and help to make modihed compounds wil enhanced properties. The relationship between these numerical properties and the activil is often described by an equation of the general form [c.711]

PW91 (Perdew, Wang 1991) a gradient corrected DFT method QCI (quadratic conhguration interaction) a correlated ah initio method QMC (quantum Monte Carlo) an explicitly correlated ah initio method QM/MM a technique in which orbital-based calculations and molecular mechanics calculations are combined into one calculation QSAR (quantitative structure-activity relationship) a technique for computing chemical properties, particularly as applied to biological activity QSPR (quantitative structure-property relationship) a technique for computing chemical properties [c.367]

Historically, QSAR has been appHed primarily to dmg molecules however, more recently. Quantitative Stmcture—Property (Toxicology) Relationships (QSP (T) R) have been elaborated by a number of researchers (117). Regardless of the context of appHcation, QSAR has its roots in the use of Linear Free Energy Relationship (LEER) methods of deriving an equation which correlates the observed effect and the stmctural features responsible for the activity. The work of Hansch in relating hydrophobicity to biological activity is exemplary as a pioneering appHcation of LEER in QSAR (118). It is interesting to note, however, that many of the early investigations in QSAR involved analysis of the relationship of hydrophobicity to activity, the nature of which relationship is more often paraboHc rather than linear (119,120). QSARs are usually best derived from a series of compounds (typically differing only at one or two substitution points) for which the activities are weU-deterrnined by a stable biological assay. A QSAR table can be estabHshed wherein the columns are assigned to activities (thejys), and the metrics or properties (the xs which can be either observable properties such as log P, hplc retention times nmr chemical shifts computed values such as shape and size descriptors dipole moments atomic charges or conformational energies. Each row represents an individual compound or conformation. Statistical relationships can then be developed from this table by means of univariate or multivariate techniques such as linear or multiple linear regression (MLR), or partial least squares (PLS). If the statistical significance of the relationship is sufficiently high, then this relationship is robust enough to be used to predict or assess the activity of untested compounds. Usually the known data is divided into two groups, a training set and a test set to estabHsh the statistical model. As a rule of thumb, there should be between 3 and 10 times the number of observations (rows) as x-variables in order to derive a model which would have predictive power and be able to minimize chance correlations. Under these circumstances, there would be some constraints against pursuing QSARs if only a few observations (molecules for which activities or properties are known) are available. Often this is the case, yet investigators have proceeded with the development of the QSAR. A discussion of the use of Partial Least Squares (PLS) as a potential means of resolution for this difficult is treated herein in the context of the CoMFA paradigm. It should also be recognized that many more molecular conformations and property descriptors than ever before can now be computed. For example, in the QSAR and Diversity modules of the Cerius2 software from Molecular Simulations, Inc. (121), there are approximately 160 shape, size, and electrostatics descriptors which can be computed. This relative abundance is in sharp contrast to the small numbers of descriptors available to early investigators, who worked diligently with classical Hammett CJs and log P values to derive LFERs (122—125). A particularly lucid account of the Hansch Approach to appHcation of LFERs in QSAR which is illustrated with numerous examples is given by the Hansch coUaborator and speciaHst in QSAR, T. Fujita (126). An example of the classical linear equation is represented by equation 7 [c.168]

Chemical Reactivity. Among other things, reactivity plays a vital role in determining fate of pollutants in the environment. A prototype computer system to estimate reactivity of systems based on stmcture, SPARC (SPARC Performs Automated Reasoning in Chemistry), is described (242). As of this writing, procedures for hydrolysis rate constants, ionisation piC uv—vis light absorption spectra, and several physical properties have been developed for SPARC. Toxicology and OtherBiological System Studies. Quantitative stmcture—relationship (QSAR) studied in the field of toxicology are extensive. These relationships range from functional group correlations to conventional group contributions to solvation energy relationships to molecular connectivity indexes. The U.S. Fish and Wildlife Service has developed an expert system that generates linear solvation energy relationship (LSER) parameters based on contaminant chemical stmcture and uses them for predicting aquatic toxicity (243). Overviews of and specific QSAR uses in environmental toxicology were presented in 1983 (244—247). An overview of QSAR studies related to dmgs acting on the central system has also been compiled (248). A historical review of research into the relationship of biological activity and aqueous solubiUty to partition coefficient as well as discussions of related QSAR-based models is available (249) as is a strategy for ranking chemical ha2ards based on QSAR techniques (250). Specific property studies on acute fish toxicity relationships can be found in the Hterature (251,252). Other studies propose QSARs for specific chemical classes, such as the organochlorine and synthetic pyrethroid insecticide study (253). [c.254]

Quantitative Structure-Activity Relationship (QSAR) Methods. These methods (62) are computer algorithms that constmct mathematical functions correlating the chemical stmcture of a compound to its biological inhibitory activity. In many cases, this has led to the prediction of a compound s biological activity after obtaining data from a series of stmcturaHy similar molecules (63,64). For one QSAR approach, the Hansch method (65—67), it is assumed that the relative potency of an analogue is related to an additive combination of terms related to its physicochemical parameters such as piQ partition coefficient, and molecular refractivity. The values of the physicochemical properties and the values of concentration necessary for eliciting a standard biological effect ate measured for a series of stmcturaHy related compounds. Next, a set of many equations with many unknowns is obtained and solved for by a least-squares multiple regression analysis. In another approach called the Free-Wilson method (68), it is assumed that the relative potency of an analogue is related to an additive combination of terms described by variables that indicate the presence or absence of functional groups. For systems that cannot be accurately described by either of these two methods alone, the methods can be combined. With hybrid Hansch—Free-Wilson equations, the effects of functional groups on biological activity that cannot be fully accounted for by the physicochemical properties alone can be quantified. Once the mathematical function has been deterrnined by one of these methods, the variables for a new but related compound can be plugged into the equation, and the equations solved for its anticipated biological potency. [c.327]

The quantitative structure-activity relationship (QSAR) approach was first introduced by Hansch and coworkers [23,24] on the basis of implications from linear free energy relationships (LFERs) in general and the Hammett equation in particular [25]. It is based upon the assumption that the difference in physicochemical properties accounts for the differences in biological activities of compounds. According to this approach, the changes in physicochemical properties that affect the biological activities of a set of congeners are of three major types electronic, steric, and hydrophobic [26]. These structural properties are often described by Hammett electronic constants [27], Verloop STERIMOL parameters [28], hydrophobic constants [27], etc. The quantitative relationships between biological activity (or chemical property) and the structural parameters could be conventionally obtained by using multiple linear regression (MLR) analysis. [c.358]

In 3D searching strategies, the relevant molecules will be found by searching the fitting 3-D molecular properties around the chirality center instead of similar functional groups, as it can be searched using 2D query structures. The facilities of ISIS/3D fulfil this ability to import and search 3D structures in CHIRBASE. The 3D structures were built with the program CORINA developed by Gasteiger s group [16] and included in TSAR [17], a fully integrated quantitative structure-activity relationship (QSAR) package. For each 2D structure, only a single conformation was stored in CHIRBASE because ISIS/3D allows conformationahy flexible substructure (CFS) searching [18]. In a CFS search, the conformational fitting is the process of rotating single bonds in 3D structures to fit the constraints of the query. Actually, searching a database of flexible conformations is quite efficient and does not require the storage of accurate models as produced from X-ray crystallographic data, as long as we always build the database with models derived from the same force field calculations and therefore control the errors. [c.106]

See pages that mention the term

**QSAR property relationship**:

**[c.169]**

Molecular modelling Principles and applications (2001) -- [ c.695 , c.702 ]