Weighting of the descriptor

Molecules can be described in many different ways, some of which are closely correlated. How then are the different descriptors to be correlated and combined into an overall description One solution, suggested from the field of QSAR is to autoscale the descriptors. This at least puts everything on an equal footing. However, one may not want to put an equal emphasis on molecular weight as opposed to the number of pharmacophores expressed. Furthermore, changes in the relative weights of the descriptors will lead to libraries of different composition. This question is as yet unresolved, and we suspect that it may have to be dealt with case by case. 

Considering the descriptors used in this paper, it is thus necessary to decide if, for example, the bioaccumulation expressed by log Kow is more or less important than acute toxicity for fish expressed by LC50. Weighting of the descriptors is a way to take into account additional external judgements. It can be seen as formalism to introduce the participation principle in the ranking procedure. 

---