QSAR Study of the Fungicidal Activity of Thiophosphorylglycinamide Derivatives. Comparison of Principal Component Analysis and Spectral Mapping Technique

Abstract

The growth inhibitory effect of 38 thiophosphorylglycinamide (TPGA) derivatives against 25 fungal species (test organisms) belonging to various taxonomic groups was determined. Principal component analysis (PCA) was used for the elucidation of similarities and dissimilarities between TPGAs and between fungi. The strength (potency) and selectivity of the fungicidal effect were separated by the spectral mapping technique (SPM). The dimensionality of the matrices of PC loadings and variables and the selectivity maps were reduced to two by the nonlinear mapping technique. PCA and SPM were compared by calculating linear relationships between the potency values and the corresponding coordinates of maps calculated by PCA and SPM. The relationship between fungicidal effect and molecular structure was elucidated by stepwise regression analysis. It was established that the coordinates of maps are orthogonal and the results of PCA and SPM are similar but not identical. The advantage of SPM is the separation of the strength and selectivity of fungicidal effect facilitating the rational design of new TPGA derivatives. Stepwise regression analysis indicated that the position of substituents in the TPGA molecule, its bulkiness and hydrophobicity exert the highest impact on both the strength and selectivity of fungicidal effect. No relationship was found between sensitivity of fungal species towards TPGAs and their taxonomic position. l IntroductionLast few decades various multivariate mathematical-statistical methods have found growing acceptance and application in quantitative structure -activity relationship studies [1,2]. While traditional regression models require a dependent variable and one or more independent ones [3, 4], multivariate methods makes possible the elucidation of the relationship between the members of any data matrix without being one the dependent variable [5]. Principal component analysis (PCA) as a versatile and easy to carry out multivariate method has been frequently employed for the elucidation of the relationship between biological activity and molecular characteristics [6]. PCA has been successfully used for improving ligand selectivity in receptor-based drug design [7], for the study of the substituent effects of meta and para benzene derivatives [8], for the characterization of amino acids [9], for the evaluation of aquatic toxicity of xenobiotics [10], and for the study of the structure-activity and structure-toxicity relationships of anti in¯ammatory drugs [11]. PCA forms clusters of variables and observations taking into consideration simultaneously the strength and selectivity of the biological effect under investigation that is PCA cannot be employed when the separation of the strength and selectivity of the effect is required. Spectral mapping technique (SPM) has been developed to overcome this dif®culty [12]. The method divides the information into two matrices using the logarithm of the original data. The ®rst one is a vector containing the potency values related to the overall eff...

Effect of the Composition of Culture Media on the Laccase Production ofLentinus edodes Strains

“…Non-linear mapping technique (NLMAP) was developed for the reduction of the dimensionality of such matrices [24]. PCA and SPM have been previously compared [25,26] and it was established that the information content of both methods differs and, therefore, their simultaneous application in QSAR studies is justified. The objectives of the investigations are the determination of the effect of the ion content of the fermentation broth on the laccase production of Lentinus edodes strains, the evaluation of the data by PCA and SPM, the comparison of the information obtained by the methods and the elucidation of the relationships between the enzymatic activity of the strains and the fermentation parameters by stepwise regression analysis [27].…”