ABSTRACT
Many approaches in 3D analysis methods rely on the so-called spectroscopic QSAR methods, sometimes also called quantitative spectroscopic data-activity relationships (QSDARs) [406]. The basic idea is that spectroscopic characteristics constitute, from an empirical point of view, a very sharp structural ngerprint of the molecular structure. But from a more theoretical point of view, they are also closely related to the eigenvalues of the corresponding Hamiltonian (e.g., normal mode frequencies in the IR, orbital energies in the UV-visible spectra), and so rešect intrinsic electronic and physicochemical properties of molecules (implicitly related to their 3D structure). Consequently, they may be considered as molecular descriptors. These descriptors are “physical observables” and not some articially calculated descriptions of the molecular structure. Such descriptors are also orientation-invariant and no alignment is necessary when using them in QSAR models [52] unlike the comparative molecular eld analysis (CoMFA)-type methods. Strictly speaking, these methods come within the 2.5D approaches since the geometrical information is not explicit, except for some interatomic distances involved in the nuclear Overhauser enhancement (NOE) effects. Finally, with these methods, no contour plots are available to specify regions where appropriate structural modications may increase biological activity.
