Fail early

Have a look at the article in Chemical and Engineering News "Drug Discovery - Filtering Out Failures Early in the Game", June5, 2000 C&EN page 63. This article is not availabe on the Internet, but more recent ones very similar to it. Search Google with "Drug Discovery - Filtering Out Failures Early in the Game".

This articles gives a pretty good overview about the different efforts of drug and software companies to develop methods to weed out early the "false negatives". "These are drug candidates that give no hint of having undesirable qualities until they are in the final stages of testing, when they're found out to have safety issues."

At Pfizer, some of the primary reasons for attrition of drug candidates are safety, pharmacology and poor ADME characteristics.

The ratios are shown below:

In terms of poor ADME characteristics, clearance is the dominant cause of attrition; the ratios documented at Pfizer are: poor absorption-20%, poor clearance-64% and poor distribution-16%.

Historically (before 1995), attrition of new chemical entities due to pharmacokinetics was almost 20%. Recently, from 1997 to 2001, pharmacokinetic-related attrition was slightly over 10%. This decrease has been attributed to better screening and identification of drug metabolism problems earlier in the discovery process.

from: Influencing Chemists to Minimize Attrition,Jae Lee, Pfizer Inc. - 2002

The methods that are mainly used for ADMET

databases:

• Reaxys
• SciFinder
• Metabolite
• Toxicity Database
• xPharm, one of the few sources with pharmacological data

and  in silico methods:

• PASS
• Metabol Expert

Here are the features of  some databases: