QSAR, or quantitative structure-activity relationship, is a mathematical model that relates numerical descriptors of a molecule's structure to a measured biological activity, allowing the activity of untested compounds to be predicted and the structural drivers of activity to be identified.

What is the difference between classical QSAR and 3D-QSAR?

Classical QSAR (such as Hansch analysis) relates activity to physicochemical or other tabulated descriptors of substituents or whole molecules; 3D-QSAR methods such as CoMFA align molecules in three dimensions and use the values of steric and electrostatic fields around them, capturing spatial structure-activity information and producing maps of where changes affect activity.

Quantitative Structure-Activity Analysis (QSAR)

Quantitative structure-activity relationship (QSAR) analysis turns the qualitative observation that structure shapes activity into a mathematical model: it relates numerical descriptors of molecular structure to a measured biological activity, so that activity can be predicted for compounds not yet tested. It is the quantitative core of structure-activity reasoning in medicinal chemistry.

Definition

A quantitative structure-activity relationship is a mathematical model that correlates numerical descriptors of chemical structure — such as physicochemical, topological, electronic, or three-dimensional field properties — with a quantitative measure of biological activity, in order to interpret structure-activity trends and predict the activity of untested compounds.

Scope

The entry covers the logic of describing molecules numerically, classical Hansch-type analysis built on physicochemical parameters, the move to three-dimensional and field-based methods, how models are built and validated, and how they are used and what limits their reliability. It treats QSAR as a modelling methodology, not as clinical guidance.

Core questions

How can chemical structure be represented numerically as descriptors?
How is a relationship between those descriptors and activity fitted and interpreted?
What do three-dimensional and field-based QSAR methods add over classical parameter-based analysis?
How is a QSAR model validated, and what defines its domain of reliable prediction?

Key concepts

Molecular descriptor
Congeneric series
Hansch analysis and substituent parameters
Free-Wilson (additive group-contribution) analysis
3D-QSAR and molecular fields
Partial least squares regression
Model validation and applicability domain
Overfitting and chance correlation

Key theories

Hansch (linear free-energy) QSAR: Within a congeneric series, biological activity can be expressed as a linear combination of physicochemical substituent parameters — characteristically a hydrophobic term together with electronic and steric terms — grounded in linear free-energy relationships, giving an interpretable and predictive model of activity.
Three-dimensional field-based QSAR (CoMFA): Comparative molecular field analysis aligns a set of molecules and computes steric and electrostatic interaction fields at grid points around them, then relates those field values to activity by partial least squares, capturing three-dimensional structure-activity information and yielding maps of where field changes affect activity.

Mechanisms

QSAR encodes each molecule as a set of descriptors — physicochemical parameters such as lipophilicity, electronic, and steric terms in classical Hansch analysis; indicator variables for the presence of groups in Free-Wilson analysis; or, in three-dimensional methods, the values of steric and electrostatic fields sampled around aligned molecules. A statistical or machine-learning method then fits the relationship between these descriptors and measured activity for a training set, producing a model that is interpreted to identify which structural features drive activity and used to predict activity for new compounds. Reliable use depends on careful validation, an honest estimate of predictive performance, and respecting the model's applicability domain — the region of chemical space the training data covers — because models can otherwise reflect chance correlation or fail outside the data they were built on.

Clinical relevance

QSAR underpins how candidate molecules are prioritised and optimised and how some property and toxicity predictions are generated in drug discovery and chemical safety assessment. The content is educational background on a modelling methodology; it describes how activity is predicted from structure and is not guidance for clinical use of any compound.

Evidence & guidelines

QSAR methodology is documented in the foundational papers that introduced parameter-based and field-based analysis and in comprehensive reviews surveying the field's development, validation practices, and best-practice expectations. These are methodological design and modelling principles rather than clinical practice guidelines; formal guidance on validation exists in the regulatory and cheminformatics literature but is summarised here only at the level of principle.

History

Quantitative SAR began in 1964 when Hansch and Fujita correlated biological activity with physicochemical substituent parameters through linear free-energy relationships, while the Free-Wilson approach offered a parallel additive group-contribution model. The compilation of partition data by Leo and Hansch supplied descriptors for this work. In 1988 Cramer and colleagues introduced comparative molecular field analysis, extending QSAR into three dimensions. The field subsequently expanded with many descriptor types and machine-learning methods, and reviews such as Cherkasov and colleagues' 2014 survey took stock of its development, validation standards, and future directions.

Debates

Predictivity, validation, and the applicability domain: How rigorously QSAR models must be validated to be trusted — including the dangers of overfitting, chance correlation, and extrapolation beyond the training set — has been a persistent concern, with the field converging on requirements for external validation and explicit applicability domains.

Key figures

Corwin Hansch
Toshio Fujita
Spencer Free
James Wilson
Richard Cramer
Alexander Tropsha

Seminal works

hansch-fujita-1964
cramer-1988
cherkasov-2014

Frequently asked questions

What is QSAR?: QSAR, or quantitative structure-activity relationship, is a mathematical model that relates numerical descriptors of a molecule's structure to a measured biological activity, allowing the activity of untested compounds to be predicted and the structural drivers of activity to be identified.
What is the difference between classical QSAR and 3D-QSAR?: Classical QSAR (such as Hansch analysis) relates activity to physicochemical or other tabulated descriptors of substituents or whole molecules; 3D-QSAR methods such as CoMFA align molecules in three dimensions and use the values of steric and electrostatic fields around them, capturing spatial structure-activity information and producing maps of where changes affect activity.