ABSTRACT
An important constraint with comparative molecular eld analysis (CoMFA) and comparative molecular similarity analysis (CoMSIA) is the need for superimposing molecules onto a common reference compound. This operation, decisive for obtaining good models, is relatively easy for a series of congeneric compounds sharing a rigid skeleton. Unfortunately, for sets of structurally varied molecules (with possibly conformational šexibility), dening an alignment is often subjective and always time consuming. This handicap led researchers to propose alignment-free systems (that is systems avoiding the superimposition of the molecules under scrutiny onto a common reference). In these methods, the values of the molecular elds are generally not directly considered but rather undergo some mathematical treatment to generate Translation, Rotation Invariant (TRI) descriptors. These approaches are closely related to 2.5D descriptors and methods but owing to the signicant common features between CoMFA, CoMSIA, and GRIND, we preferred to deal with autocorrelation in a separate chapter.
