Consensus scoring scheme

Fig. 3. MapMaker is a combinatorial library optimization tool developed at ArQule, currently used in the design of screening libraries. The system is represented as a black box, since the details of operation are hidden by a web interface, to which chemists provide a reaction scheme, lists of candidate reagents, and the number of reagents desired for each dimension of a library, and from which they retrieve lists of reagents that encode for the optimized library. Internally, MapMaker enumerates the full virtual array, calculates the desired properties and coordinates in chemical space for the virtual compounds, and performs the optimization using a genetic algorithm. Consensus scoring following property calculation allows the system to optimize around an arbitrary number of computed properties, which are determined at run-time.

In a farsighted move in 1989, the European Union laboratory IRMM started a series of interlaboratory comparisons to provide objective evidence for the degree of equivalence and the quality of chemical measurements by comparing a participant s measurement results with external certified reference values (IRMM 2006). At the time most proficiency testing schemes used consensus results for the mean and standard deviation to derive z scores. With the IMEP-1 analysis of lithium in serum, the world was alerted to the problem of lack of accuracy in analytical measurements. The data of the first IMEP-1 trial are replotted in figure 5.6 notice that the apparent outlier was the only laboratory to come close to the assigned value. 

Zeta (z) scores are commonly used as a performance indicator in PT schemes [37, 61, 62]. Sometimes, a consensus value derived from participants results can serve as reference and the performance criteria, sometimes called target value, equals one standard deviation. This means that 95 percent of the participants would automatically report satisfactory results. In IMEP a z-score is meant as ... 

---