Study And Application Of Quantitative Spectrum-Activity Relationship

Serving as a common technique for molecular design, quantitative structure-activity/ property relationship (QSAR/QSPR) plays an important role in research and development of new drugs. QSAR has greatly improved many subjects such as medicinal chemistry. When to process spectral data, it is referred to quantitative structure-spectrum relationship (QSSR). Due spectral complexity and diversity, it is quite difficult to be simulated correctly. In this context, some spectral behaviors of organic compounds and bioactive molecules are deeply examined by several methods of new molecular structural characterization (MSC). Major works are metioned as follows.â‘ For two-dimensinal molecular structures, atomic electronegativity interaction vector (AEIV) and atomic hybridation state index (AHSI), are developed to demonstrate atomic microscopic environment and hybridation states in molecules. These descriptors are used to characterize various equivalent resonant carbons of 48 pyran and 25 furan glycoses, multiple linear regreesion (MLR) is constructed to simulate nuclear magnetic resonance (NMR) and predict 13C chemical shifts with modeling estimation r and cross validation q correlation coefficient both above 0.9 by strict statistical diagnosis.â‘¡As novel molecular coding, atom-pair holograph (APH) is obtained by defining 36 atomic fragments for different organic compounds and by constructing multilevel pair-frequency matrix according to their occurrence with different bond-lengths. This method exhibits several benefits, such as simple calculation, easy operation and definite meaning, to fetch complicated molecular information, etc. Therefore, it is appropriate for quantitative structure-retention relationship (QSRR) of medicines and biomolecules. APH is utilized to precise prediction of reversed-phase liquid chromatogram (RPLC) retentions of both 33 purines and 24 steroids. Model-estimated and cross-validated multiple correlation coefficients are respectively obtained as r²=0.990, 0.893 and q²=0.977, 0.897, with predictive ability r²_pred=0.941 by partial least-square (PLS).â‘¢Furthermore APH is also utilized for quantitative structure-spectrometry relationship (QSSR) of a large-scale ion mobility spectrometry collision cross section database comprising 819 peptides collected from references. It is confirmed that APH is outstandingly related with collision cross sections, while involving in partially nonlinear factors for few polypeptides. Deeply testing estimated stabilities and generalized abilities by both internal and external exams, the model is deemed to assist in quantitative computer-aided predictions for peptide collision cross sections.â‘£on structural representation, a novel generalized correlative index (GCI) is derived from generalized correlative function (GCF), property correlative parameters (PCP) and distance-relational function (DRF). GCI is employed for retention behaviors of polychlorinated 41 dibenzo-p-dioxins, 115 dibenzofurans, 62 naphthalenes and 210 biphenyls with good QSRR models having both r and q above 0.98. GCI is thus deemed to be adaptable to diverse molecular systems.⑤As novel electrotopological descriptors, molecular electronegativity-interaction vector (MEIV) has been developed to characterize structures of 420 singly protonated peptides. In order to find linkage between structure and collision cross section, three QSPR models were built and proved with excellent fitness and good predictability for both internal and external validations with both r and q above 0.95. Satisfactory results show that MEIV correlates well with collision cross section, mainly linear and somewhat nonlinear relationship.â‘¥By mapping intrinsic interatomic correlations into a certain coordinates system according to a given reasonable sampling interval by radical distribution function, molecular graphic fingerprint (MoGF) is proposed for novel structural characterization related to ion mobility spectrometry (IMS) applications mainly focused on intricate drug/biological systems. With abundant contents of structural information, a characteristic graph is obtained possessing of great merits in easy calculation, independent of experiments, explicit structural meanings and intuitive expressions, etc. MoGF is used for collision cross sections of 579 singly-protonated peptides, the constructed robust and predictable PLS models are rigorously subject to both internal and external validations, with statistics on both training and test sets as r²=0.991, q²=0.990, RMSEE=5.526, RMSCV=5.572, q_ext²= r_ext²= r_0,ext²= r_0,ext^'2=0.990, k=1.003, k′=0.996 and RMSEP=5.561, respectively.