QSRR Models Research Articles

Reversed-phase HPLC retention data in correlation studies with in lipophilicity molecular descriptors of carotenoids

Influence of chemical structure on the lipophilicity of isolated free carotenoids from paprika oleoresin has been studied by QSRR approach (Quantitative structure-retention relationship). The chromatographic behavior of these compounds was investigated by using reversed phase high-pressure liquid chromatography (RP HPLC). Retention mechanism has been determined using the following mobile phase: acetone -water, on reversed - phase column (SB-C18). A variety of lipophilicity parameters (logP) were calculated by use of different software products. On the basis of correlations, the nonlinear structure-activity models were derived between the retention constants, tr (retention time of investigation compounds), and logP values. Five high quality QSRR models were found to have a good predictive ability and close agreement between experimental and predicted values. The study showed that the retention constants can be used as a measure of lipophilicity of investigated compounds at a high significant level.

QSRR Study on Flavor Compounds of Diverse Structures on Different Columns with the Help of New Chemometric Methods

A quantitative structure–retention relationship study was performed for 286 flavor compounds with highly structural diversity on different stationary phases of different polarities. Firstly, the structure–retention relationship was statistically explored using constitutional, topological, molecular properties, charge descriptors and quantum chemical descriptors. Some recently developed chemometric methods, such as the ones for robust analysis and variable selection were firstly employed to predict accurately the gas chromatographic retention indices. The stability and validity of models have been tested by internal and external validation, and good robustness and predictive ability were obtained. The resulting QSRR models were well correlated, with the square of correlation coefficient for cross validation, Q 2, values of 0.9847, 0.9838, 0.9745 and 0.9646 on stationary phase DB5, OV101, OV17 and C20M, respectively. The molecular properties known to be relevant for GC retention index, such as molecular size, branching, electron density distribution and hydrogen bond effect were well covered by generated descriptors. The descriptors used in models on four stationary phases were compared, and some reasonable explanations about gas chromatographic retention mechanism were obtained. The developed model may be useful for the prediction of flavor compounds while experimental data are unavailable.

Prediction of Gas Chromatographic Retention Indices of Polychlorinated Dibenzothiophenes and Polychlorinated Dibenzothiophene Sulfones by Electrotopological State Index

Structures of 135 polychlorinated dibenzothiophenes (PCDTs) and 135 polychlorinated dibenzothiophene sulfones (PCDTO2s) were characterized using electrotopological state index (ETSI) for atom types. Quantitative linear relationship models were built between gas chromatographic retention index (RI) from DB-5 GC column and ETSI descriptors were established for PCDT and PCDTO2. The models derived from the variable selection and modeling based on prediction (VSMP) technique present correlation coefficient (r) of 0.9939 for PCDT and of 0.9729 for PCDTO2, leave-one-out cross-validation coefficient (q) of 0.9921 for PCDT and of 0.9692 for PCDTO2, respectively. In order to validate the stability and predictive power of the models, 17 PCDT and PCDTO2 were used as training sets to build QSRR models. The models 1118 化 学 学 报 Vol. 70, 2012 present r of 0.9959 for PCDT and of 0.9783 for PCDTO2, q of 0.9921 for PCDT and of 0.9740 for PCDTO2. The results demonstrate the robustness of the QSRR models. The models were used to predict the RI of test set with 8 compounds and predict set with 110 compounds, and the result of test set with 8 compounds show that the training set has a good predictive ability.

Comparison of quantitative structure–retention relationship models on four stationary phases with different polarity for a diverse set of flavor compounds

A quantitative structure–retention relationship study was performed for 656 flavor compounds with highly structural diversity on four stationary phases of different polarities, using topological, constitutional, quantum chemical and geometrical descriptors. Statistical methods were employed to find an informative subset that can accurately predict the gas chromatographic retention indices (RIs). Multivariable linear regression (MLR) was used to map the descriptors to the RIs. The stability and validity of models have been tested by internal and external validation, and good stability and predictive ability were obtained. The resulting QSRR models were well-correlated, with the square of correlation coefficients for cross validation, Q 2, values of 0.9595, 0.9528, 0.9595 and 0.9223 on stationary phase OV101, DB5, OV17 and C20M, respectively. The molecular properties known to be relevant for GC retention index, such as molecular size, branching, electron density distribution and hydrogen bond effect were well covered by generated descriptors. The descriptors used in models on four stationary phases were compared, and some reasonable explanations about gas chromatographic retention mechanism were obtained. The model may be useful for the prediction of flavor compounds while experimental data is unavailable.

QSRR Models to Predict Retention Indices of Cyclic Compounds of Essential Oils

Genetic algorithm and multiple linear regression (GA-MLR), partial least square (GA-PLS) and Levenberg–Marquardt artificial neural network (L–M ANN) techniques were used to investigate the correlation between retention indices (RI) and descriptors for 40 cyclic compounds in rosemary and sage essential oil. Descriptors of the GA-MLR model were selected as inputs in the L–M ANN model. The R2 and RMSE for calibration, prediction and validation sets are (0.985, 0.985, 980), (0.013, 0.009, 0.010), respectively. This indicates that L–M ANN can be used as an alternative modeling tool for quantitative structure–property/retention relationship (QSPR/QSRR) studies. This is the first research on the QSRR of the essential oil compounds against the RI using the L–M ANN model.

Quantitative structure-retention relationships study of the retention data of 5,5-disubstituted hydantoins

Summary The behavior of 5,5-disubstituted hydantoin derivatives has been studied using four chromatographic systems, two normal-phase (silica gel stationary phase with ethyl acetate–toluene and acetonitrile–toluene mobile phases) and two reversed-phase (RP-18 systems with methanol–water and acetonitrile–water mobile phases). The effect of the structures of the derivatives on their retention in both normal and reversed-phase modes was investigated by use of QSRR and molecular descriptors. By means of multiple linear regressions results were interpreted in terms of understandable and physically meaningful variables. Among six models proposed, the two best were selected and tested for prediction of retention – one for normal-phase chromatography (NPC) and the other for reversed-phase chromatography (RPC). The most informative properties in the best NPC model, with similar effects on retention, are log P and valence connectivity indices. The informative properties in the best RPC model were log P and molecular weight, but the statistics was much worse than for NPC. Cross-validation indicated the best models are reliable QSRR models.

Advanced QSRR modeling of peptides behavior in RPLC

In QSRR the retention is modeled as a function of structural or molecular descriptors. Since the structural datasets can be very large a selection of informative variables is often required. But beside the question which subset of variables (descriptors) produces optimum predictions one should answer the question: can good prediction be used in the QSRR community even if the physical meaning of applied descriptors is hard to interpret? The main focus in this paper is put on different modeling methodologies applied and molecular descriptors used in the QSRR approaches. Besides the widely used multiple linear regression (MLR), these methodologies include partial least squares (PLS), uninformative variable elimination partial least squares (UVE-PLS), genetic algorithms (GA) prior to MLR or PLS. The comparison will focus on the predictive performance but also on the descriptors found to be most important for the chromatographic retention prediction of peptides. The results of this study showed that stepwise-MLR and UVE-PLS are producing better predictions than the rest of the studied methodologies. From the variables selected by various methodologies one can see that the important information for the retention mechanism of RPLC was given by 2D-, 3D-descriptors and descriptors from the empirical QSRR equations, which bring the information about hydrogen-bonding properties, molecular size, and complexity. Overall, for the considered data set the empirical QSRR models were predicting the peptides retention best.

Comparison of physicochemical and gas chromatographic polarity measures for simple organic compounds

The comparison of different polarity measures (parameters, descriptors, variables, scales, etc.) indicates that evaluation of interrelations between these measures is important for better understanding and interpretation of chemical and/or analytical data, especially for chromatographic separation. The best linear correlation between gas chromatographic and non-chromatographic polarity descriptors is revealed for the first time: this pair of variables is the difference of gas chromatographic retention indices on standard polar and non-polar phases as well as the difference between non-dimensional indices of boiling points (known in chromatography since mid-1980s as dispersion indices) and indices of molar refractions. The correlation helps chromatographers to find preferable chemical variables (features) to understand better the separation phenomena and to find better correlations in QSRR models. Principal component analysis (PCA) of ten frequently applied polarity measures shows their similarity and, at the same time, it shows the absence of anomalies within the set of simple organic molecules. A novel ranking method for ten polarity parameters points out that the two most informative polarity measures are (i) the non-dimensional index for boiling point and (ii) the difference in chromatographic retention indices on standard polar and non-polar stationary phases. On the other hand, the hydrophobicity parameter, log P, sometimes considered as polarity parameter in HPLC seems to be the worst one in description of “polarity” in gas chromatography. Surprisingly, such polarity measures like dipole moment and permittivity used often in organic chemistry does not provide the best correlation with gas chromatographic polarity measures.

Evaluation of a generalized use of the log Sum(k+1)AA descriptor in a QSRR model to predict peptide retention on RPLC systems

At the current state of knowledge, the rational optimization of the chromatographic separation of peptides, as well as the identification of proteins in proteomics are challenges for analytical chemists. In this paper the generalized applicability of a recently derived descriptor log Sum(k+1)AA in a QSRR equation to model peptide retention in RP-LC systems was evaluated. For that purpose, two sets of peptides analyzed on dissimilar RP-LC systems were considered. A first set of 28 peptides was measured on 17 columns/systems, while a second of 70 peptides was eluted on four. The aim of this work was to confirm the usefulness of the partly experimental log Sum(k+1)AA descriptor for the prediction of peptides retention compared to the initially applied, fully experimental log SumAA descriptor. The verification of the predictive abilities of both QSRR models, applying either the initial or the alternative descriptor, was done by using the leave-one-out and leave-three-out cross-validation procedures. The results seem to demonstrate that the QSRR model with log Sum(k+1)AA, for which the retention measurement of only seven out of 20 existing amino acids is necessary, possesses similar or in some cases even better predictive abilities than that containing log SumAA.

Quantitative structure–retention relationship modelling of esters on stationary phases of different polarity

The semi-empirical electrotopological index, I SET, used for QSRR models, was developed and optimized to describe the chromatographic retention of saturated esters on the five different stationary phases (SE-30, OV-7, DC-710, OV-225 and XE-60). The simple linear regressions between the retention indices and the proposed index were of good statistical quality, high internal stability and good predictive ability for external groups, especially for the stationary phase with low polarity, showing that the specific molecular interactions occur on highly polar phases. For the esters, the interactions between the molecules and the stationary phase are slowly increased relative to hydrocarbons due to the charge redistribution that occurs in the presence of the heteroatom. These facts were included in the calculation of I SET through a small increase in the SET i values for heteroatoms and the carbon atoms attached to them. The increase in the SET i values originates from the dipole moment of the whole molecule and an equivalent local dipole moment related to the net charges of the atoms belonging to the functional group and the carbon atoms attached to them. The polarity of the stationary phases, indicated by the retention polarity ( P R) given by Tarján, is reflected in the intercept of the equations obtained for each stationary phase. Thus, a single combined QSRR model was generated with a satisfactory predictive quality, including a parameter that represents the polarity retention of all stationary phases studied.

The molecular descriptor log Sum AA and its alternatives in QSRR models to predict the retention of peptides

The use of the experimental molecular descriptor log Sum AA and some possible alternatives were evaluated in the QSRR analysis of peptides. To quantitatively characterize the structure of analytes in a previously proposed QSRR the following three structural descriptors were applied: the logarithm of the sum of gradient retention times of the amino acids composing the individual peptide, log Sum AA; the logarithm of the peptide’s van der Waals volume, log VDW Vol; and the logarithm of its theoretically calculated n-octanol–water partition coefficient, clog P. Taking into consideration that most amino acids were hardly retained in the different RP-HPLC systems on which the peptides retention was measured, the contribution of most amino acids to the log Sum AA descriptor is rather constant. Therefore, to enlarge the variability of the descriptor and the amino acids contributions for a given series of peptides, in a first instance, it was evaluated whether, by changing the chromatographic conditions, the retention differences between the amino acids could be increased, while maintaining their mutual selectivity. It was not evident to find such conditions. Secondly, it was also investigated whether the experimental descriptor log Sum AA can be replaced by a theoretical, either based on a simple or on a weighted counting of the amino acids composing the peptide. The weighting factor for the retained amino acids was determined by their experimental gradient retention times measured on different systems. The predictive abilities of the new QSRR models (applying the alternative descriptors) were assessed using the leave-one-out cross-validation procedure and compared to that of the initial model. Finally, a descriptor was defined for which the retention measurement of only a limited number of amino acids is required. It resulted in QSRR models with similar predictive properties as those with log Sum AA, but with a reduced workload.

QSPR study of GC retention indices for saturated esters on seven stationary phases based on novel topological indices

The novel topological indices, polarizability effect index (PEI), odd–even index (OEI) and steric effect index (SV ij ), previously developed in our team, were extended to predict the gas chromatographic retention indices (RI) of saturated esters on seven stationary phases (SE-30, OV-7, DC-710, OV-25, XE-60, OV-225 and Silar-5CP). The sets of molecular descriptors were derived directly from the structure of the compounds based on graph theory. Multiple linear regression (MLR) models between the RI and the topological indices were established for each stationary phase with a correlation coefficient between 0.9989 and 0.9977 and a leave-one-out cross-validation correlation coefficient between 0.9988 and 0.9975. The average prediction errors over seven phases are within the range of 0.5–0.7%. Statistical analysis showed that the polarizability effect, the molecular size and the branching make dominant contributions to RI of ester compounds, but steric effect also provides separate contribution. The models with topological indices were compared with recently proposed QSRR models of the similar data. It is found that the present indices produce better correlations with Kováts retention indices than the previous topological indices. The significant improvement demonstrates the efficiency of the current study in the quantitative structure retention indices correlations of complex compounds with polar functional group.

Novel generalized correlative index used to study the quantitative structure-retention relationship on several persistent organic pollutants

Ideas regarding to facing to users and self-regulation are introduced into molecular structural representation. Based on molecular topological structures and branch-connecting types, a novel molecular representing method as generalized correlative index (GCI) has been derived via some definitions as generalized correlative function (GCF), property correlative parameters (PCP), and distance-relational function (DRF). Applying GCI into retention behavior studies for several systems as 115 polychlorinated dibenzofurans, 41 polychlorinated dibenzo-p-dioxins, 62 polychlorinated naphthalenes, and 210 polychlorinated biphenyls, the resulting QSRR models have been found satisfying with all the r (correlative coefficient) and q (cross-validated correlative coefficient) above 0.98. Adaptable to diverse molecular properties, GCI is thus deemed to be superior to represent molecular structures.

Preliminary study of the retention behavior for different compounds using cryogenic chromatography at different initial temperatures

The Kováts indices ( I) of 24 volatile organic compounds were measured at the different initial temperatures by a laboratory-made cryogenic chromatography. The results show that the correlations of the Kováts indices and the temperature are nonlinear at low initial temperature. Seven QSRR models were built using the heuristic method at different initial temperatures by calculating the quantitative chemical parameters of 24 organic compounds to study the relationship between the retention behavior and the molecular parameters. By studying the relationship between carbon and retention behavior, the capacity factor, and resolution of selected compounds, the contribution of gas–liquid interfacial adsorption process, which influences the retention behavior of solutes, is becoming more and more important with the decrease of the initial temperature.

Quantitative structure–retention relationships for ionic and non‐ionic compounds in biopartitioning micellar chromatography

Quantitative structure-retention relationships, QSRRs, represent a powerful tool in chromatography. The objectives of QSRR studies are to predict the chromatographic retention behaviour of solutes based on their structural properties, to elucidate retention mechanisms, to optimize the separation of complex mixtures or to prepare experimental designs. In this paper, using the retention factors of 151 structurally unrelated solutes that cover a wide range of hydrophobicity, molecular size, hydrogen bonding properties and ionization degrees obtained in biopartitioning micellar chromatography (BMC) at different Brij35 micellar concentrations, several multivariate QSRR models are tested. It is demonstrated that the chromatographic retention of any molecule in BMC, independently of its family, can be adequately described by its hydrophobicity (expressed as log P) and its anionic and cationic total molar charge (expressed as alpha(A) and alpha(B)).