Quantitative Structure-Activity Relationship
(QSAR) studies are widely used in drug design, in particular, for hit-to-lead and lead optimizations.
The problem can be easily investigated in ChemoSoft by a complex set of tools for data exploration
called "Explore data".

The kit contains both instruments of classical statistics (multiple linear and more advanced polynomial regression) and tools of state-of-the-art methodologies, usually regarded as "pattern recognition" techniques. The latter includes Principal Component Analysis for the reduction of descriptor space dimensionality and the most powerful ChemoSoft functionality in this area - Artificial Neural Network (ANN). The last opportunity includes multilayer perceptron. Data mining by various correlation methods and non-linear mapping of objects adds even more value to and enlarges the possibilities of the kit.

Linear regression.	Polynomial regression.

PCA: scores.	PCA: loadings.

Artificial Neural Network.

Correlation techniques: bivariate statistics.	Correlation techniques: map of variables.

The non-linear mapping of molecules can also be considered as an appreciable supplement to ChemoSoft's diversity tools represented by different algorithms of structural diversity (including the unique Diversity of Heterocycles) of molecules and "supermolecules" (Diversity of Plates) by the diversity of descriptor space. Thus, the feature is described in the "Diversity\Similarity" section in more detail.

Non-linear map of molecules.

Convenient pre-processing utilities as normalization of data, tracing missed data, make the set of QSAR (or Structure-Property Correlation - SPC in general) instruments complete and ready to raise your leads one step up to drugs.

An example of pre-processing utilities: simple statistics.