Correlation methods, molecular similarity

A fairly good correlation between the storage modulus and the relaxation rate (1/T2) was observed. Both methods gave similar values for the mean molecular mass of network... 

F1F and DFT methods have been used by Smith et al.205 to calculate the dipole polarizabilities, a, of the all trans polyenes up to CigH2o, and polyacenes to CigH12 and their molecular ions. The static values and the dynamic response at 800 nm have been calculated. For the smaller molecules the results have been compared with those of correlated methods. It is found that the uncorrelated results are about 20% higher than those of the correlated methods for the neutral molecules but are very similar for the molecular ions. General conclusions about the scaling and frequency dependence of a values obtained by simpler methods are drawn. 

To exemplify a molecular similarity method, we employed here a 3D shape-based molecular similarity approach using OpenEye scientific software (OpenEye). A set of 27 molecules (Amoore, 1971) were compared to benzaldehyde (query molecule). The representation used here is based on the volume of each molecule. A conformational ensemble is built for the molecules in the database, whereas the conformation of the query remains fixed (the chemical nature of benzaldehyde does not entail different conformers, though in many cases the conformation of the query molecules might be complex and crucial). After the conformers of each molecule in the data set are built, each one of them is compared with the query and a similarity value is computed. For the particular program employed here (ROCS), the similarity is quantified as a score formed by two terms, one takes into account the chemical nature of the molecules while the other relies on molecular shape, such score is referred to as combo score. The maximum similarity value is 2 which can only be obtained from the comparison of a molecule with itself in the exact same conformation (perfect match). The normalized values (from 0 to 1) for the odor and combo score similarities are compared in the graph shown in Fig. 2.4. As can be observed, as the combo score increases, the odor similarity to benzaldehyde also increases. This correlation shows that part of the odor similarity was captured by the molecular... 

Jain, A.N. (2000) Morphological similarity a 3D molecular similarity method correlated with protein—ligand recognition. /. Comput. Aid. Mol. Des., 14, 199-213. 

For SCF methods, p represents elements of the one- and two-particle density matrices. In correlated methods, the one-particle part of p is the sum of the aaual reduced density and a contribution that is proportional to the derivative of the energy with respect to orbital rotations. For the MCSCF method, the orbitals are variationally optimum, and this latter term vanishes. Similarly, for FCI there is no orbital contribution. However, it is required for other Cl, CC, and MBPT correlated methods, because the energy in these approaches is not stationary with respect to first-order changes in the molecular orbitals. This... 

Jain AN. Morphological similarity A 3d molecular similarity method correlated with protein-ligand recognition. J Comput Aided Mol Des 2000 14(2) 199-213. 

An issue that comes into both similarity fnsion and consensus methods is the independence of the different similarity measnres employed in a given procedure. This issue also occurs in activity landscape modeling (see Section 15.6.1) when fnsion or related methods are used to estimate molecular similarities [144, 146]. In snch cases, the qnestion typically arises when two methods produce highly correlated values as to whether the set of values from one of them should be removed from the fusion process. The answer seems to be yes, althongh most of the time in cheminfor-matics it is ignored. But should it be ... 

How does this example apply to the use of multiple similarity methods Each of the similarity methods can be considered to be equivalent to an independent judge, since none of the values produced by the other methods have an explicit impact on the value produced by a given method. This may not always be the case, for example, if two methods use MACCS key fingerprints, but one uses the Tanimoto (Jacard) and the other a closely related similarity function (see Table 15.3). As shown by Gower [76], some molecular similarity functions are monotonically related. Thus, comparisons of these functions based on the same molecular representation will produced linear correlations of the values computed by the two functionally similarity functions. Hence, only one of the functions should be used. 

---