Drug Bioavailability, Distribution and Clearance Prediction
William Bains · Encyclopedia of Molecular Cell Biology and Molecular Medicine · 2006
Prediction of how drugs are absorbed, metabolized, distributed round the body, and excreted is increasingly important to the design of new medicines. However, it is too complex to model rigorously. Prediction algorithms are primarily based on databases of known examples, which are mined for simple rules, more complex mathematical algorithms, or rules sets. Many modeling techniques have been applied, and many ways of describing a molecule to the computer have been tested. I discuss the application of these methods for predicting whether a drug will be absorbed through the intestines, including whether it will be soluble, transported in or out by cell membrane transporters, and whether it will be metabolized, and for predicting penetration into the brain. Advanced statistical and machine learning techniques can generate good algorithms for approximating these quantities. However, predictions are not yet good enough to integrate into a “virtual human”. The main limitation is that the algorithms need more experimental data to encompass the complexity and variability of biological systems, as well as better, more biologically relevant ways of describing the molecules to the computer.