This section of the course deals with a number of methods that use statistics to identify correlations between some observed property (usually a biological activity) and a set of molecular properties. Common applications include defining a equation that will reproduce the measured activity for compounds within a particular data set, and then using this to search through a chemical library (which could have been constructed for a completely different purpose) to identify new, active, compounds.

General Discussions

• A historical perspective on QSAR (here)
  
• A good general review (here)
• errors associated with high throughput screening data (here)
  

Descriptors

• Use of PCA to identify descriptors for the amino acids (here)
• Use of Genetic Algorithms to identify descriptor sets, applied to the prediction of anitfungal activity (here)
  

papers used in discussing QSAR methods

• Linear regression methods, applied to passage of drugs into breast milk (here)
• Neural Networks, applied to antifungal activity (here)
  
• 3D QSAR: development of a pharmocaphore model to identify transporters for various drugs (here)

QSAR Applications

• Antifungal activity (here)
• Nonsteroidal Anti-Inflamatory Drugs (here)
  
• Anti-tuberculosis agents (P.M. Sivakumar et al., Chem. Pharm. Bull. 2007, 5544—49)
  
• Melanoma toxicity (here)
  
• Anti-epileptic sulfamides (here)
  
• Drug transfer into human breast milk (here)
  
• ADME/tox evaluation (here)