•Modeling was performed for predicting the adsorption of organics by MWNTs and the dispersibility of SWNT in organic solvents.•Norm indexes descriptors were developed based on molecular structure.•Performances of the models were verified by several validation methods.•Norm indexes have potential profiles for use in nanomaterials.Carbon nanotubes (CNTs) have important role in ecological environment owing to their ability of the adsorption of organic contaminants which might greatly affect the dispersibility of CNTs in aquatic environments. Thus in this work, quantitative nanostructure-property relationship modeling studies were performed with the norm indexes descriptors our group proposed to predict the adsorption data (represented by logK∞ and logKSA) of organic compounds by multi-walled CNTs and the dispersibility (represented by logCmax) of single-walled CNT in various organic solvents. Calculation results showed that the three models could provide accurate and satisfactory predictions with the squared correction coefficient for the training set and the test set of 0.9500 and 0.9792 for logK∞, 0.9258 and 0.9770 for logKSA, 0.9511 and 0.9956 for logCmax respectively. Validation results containing cross validation, Y-randomized test and applicability domain analysis together with the comparison with other works demonstrated that our models were stable, robust and reliable. These satisfactory results showed that the norm indexes descriptors our group proposed might have extensive and promising applications in nanotechnology.Download high-res image (184KB)Download full-size image

•A norm index could apply in a large field is proposed.•The QSPR model built in this work with the statistical results Rtraining2, Rtest2 and MAE of 0.880, 0.933 and 0.295, respectively.•This new method has better precision than other models of literatures.Fullerene (C60) is playing an important role in nanomaterial. The solubility of C60 in various organic solvents (logS) is one of the most considerable properties, which markedly influences the extraction, purification and the subsequent organic functionalization for C60. A quantitative structure-property relationship (QSPR) model was developed to predict logS of C60 in diverse organic solvents with our previously proposed norm indexes. Results suggested that this model could give satisfactory prediction effect. The squared correlation coefficient for training set (Rtraining2), test set (Rtest2), the leave-one-out cross validation (Q2) and the overall mean absolute error (MAE) were 0.8798, 0.9327, 0.8933 and 0.2951, respectively. Results of comparison with other references' methods indicated that this model could effectively improve the accuracy and predictive ability for predicting logS of C60 in organic solvents with diverse structures.

Arylpiperazine derivatives are promising 5-hydroxytryptamine (5-HT) receptor ligands which can inhibit serotonin reuptake effectively. In this work, some norm index descriptors were proposed and further utilized to develop a model for predicting 5-HT1A receptor affinity (pKi) of 88 arylpiperazine derivatives. Results showed that this new model could provide satisfactory predictions with the square of the correction coefficient (R2) of 0.8891 and the squared correlation coefficient of cross-validation (Q2) of 0.8082, respectively. In addition, the applicability domain of this model was validated by using the leverage approach and results which suggested potential large scale for further utilization of this model. The results of statistical values and validation tests demonstrated that our proposed norm index based model could be successfully applied for predicting the affinity 5-HT1A receptor ligands of arylpiperazine derivatives.

Ionic liquids have attracted much attention in past decade for their unique properties. The density is an important property in industrial application, however the density data are relatively scarce compared with the great number of ILs. The quantitative structure property relationship (QSPR) and group contribution method (GCM) have been used for predicting ILs density. However, the accuracy of QSPR is not as good as that of GCM. In this work, a desirable QSPR model was developed to estimate ILs density. The general topological indexes (TIs) proposed by our research group were used to develop the QSPR model, which was on the basis of 5948 experimental data points for 188 ILs. The collected data are in the range of temperature (253.15–473.15) K and pressure (0.1–250) MPa. The correlation coefficient (R2) and the overall average absolute deviation are 0.998 and 0.422 %, respectively.

•A general topological index was proposed based on atom characters and position.•A QSPR model was developed to predict the glass transition temperatures of ILs.•A topological index was generated from anion and cation.Based on the general topological index (TI) proposed in our previous work, a Quantitative Structure–Property Relationship (QSPR) model was developed to predict the glass transition temperatures of ionic liquids (ILs). Because ILs are a class of molten salts which are composed entirely of cations and anions, in general, the descriptors for ILs are calculated from cations and anions separately and the interaction between them is neglected. In this study, except the two sets of TIs generated from cation and anion, a third TI was used to depict the interaction of anion and cation. The QSPR model is based on five kinds of ILs, which are imidazolium (Im), pyridinium (Py), ammonium (Am), sulfonium (Su), triazolium (Tr). The regression coefficient (R2) and the overall average absolute relative deviation (AARD) are 0.894 and 3.32%, respectively.

•A general topological index was proposed based on atom characters and position.•The topological index was extended for predicting melting points of ionic liquids.•A topological index was generated from anion and cation.A Quantitative Structure Property Relationship (QSPR) model was developed to predict the melting points of ionic liquids (ILs) with diverse classes of cations and anions. The QSPR model was based on the general topological index (TI) proposed in our previous work. The TI was successfully used for the prediction of the decomposition temperature of ILs and the toxicity of ILs in acetylcholine esterase and Leukemia Rat Cell Line. ILs are a class of molten salts which are composed entirely of cations and anions, therefore the descriptors for ILs are generally calculated from cations and anions separately and the interaction between them is neglected. In this study, besides the two sets of TIs generated from cations and anions, a third TI was used to depict the interaction of anions and cations. The QSPR model is on the base of eight kinds of ILs, which are imidazolium, benzimidazolium, pyridinium, pyrrolidinium, ammonium, sulfonium, triazolium and guanidinium. The regression coefficient (R2) and the overall average absolute deviation (AAD) are 0.778 and 7.20%, respectively.

On the basis of the new topological index (TI) proposed in our previous work, a multiple linear regression (MLR) model was developed for predicting the toxicity of ionic liquids (ILs) in Leukemia Rat Cell Line (log EC50 IPC-81). The TI is derived from atom characters (e.g., atom radius, atom electronegativity, etc.) and atom position in the hydrogen-suppressed molecule structure. Because ILs are composed entirely of cations and anions, the TIs are calculated from cation and anion, respectively. A third TI was also proposed to depict the interaction of anion and cation. The toxicity of 173 ILs, which are based on imidazolium (Im), pyridinium (Py), pyrrolidinium (Pyr), ammonium (Am), phosphonium (Ph), quinolinium (Qu), piperidinium (Pi), and morpholinium (Mo), was calculated by the model. The regression coefficient (R2) and the overall average absolute error (AAE) are 0.938 and 0.226, respectively.

In this work a new topological index (TI) was proposed based on atom characteristics (e.g., atom radius, atom electronegativity, etc.) and atom positions in the hydrogen-suppressed molecule structure. Using the TIs, a multiple linear regression (MLR) model was developed for predicting the decomposition temperature (Td) of 158 ionic liquids (ILs), which are based on imidazolium, pyridinium, pyrrolidinium, ammonium, phosphonium, sulfonium, and guanidinium. Because ILs are a class of molten salts which are composed entirely of cations and anions, in general, the descriptors for ILs are calculated from cations and anions separately, and the interaction between them is neglected. In this study, except for the two sets of TIs generated from cations and anions, a third TI was proposed to depict the interaction of anions and cations. The regression coefficient (R2) and the overall average absolute deviation (AAD) are 0.893 and 3.07 %, respectively.

A new topological index (TI) was proposed based on atom characters (e.g., atom radius, atom electronegativity, etc.) and atom positions in the hydrogen-suppressed molecule structure in our previous work. In this work, the TI was used for predicting the toxicity of ILs in acetylcholin esterase (log EC50 AChE) by the multiple linear regression (MLR) method. For ILs composed entirely of cations and anions, the TIs are calculated from cations and anions, respectively. The 221 ILs used in the MLR model are based on imidazolium (Im), pyridinium (Pyi), pyrrolidinium (Pyo), ammonium (Am), phosphonium (Ph), quinolinium (Qu), piperidinium (Pi), and morpholinium (Mo). The regression coefficient (R2) and the overall average absolute error (AAE) are 0.877 and 0.153, respectively.

A new universal method was proposed for the prediction of properties of organic compounds, such as critical properties, normal boiling point, and the enthalpy of vaporization. In this study, the positional distributive contribution method is further extended for the prediction of the acentric factor ω of a variety of pure organic compounds. Comparison results between experimental and calculated data indicate that the new model could provide very satisfactory results. The overall average absolute error for the ω prediction of 477 organic compounds is 0.0252 with 5.72 % mean absolute relative deviation, which is compared to 0.0569 and 14.58 % with the Constantinou and Gani method. Also, a good prediction of the proposed method shown in our previous works and this work suggests that it is possible to use the same universal formula to predict not only Tc, Pc, Vc, Zc, Tb, Tm, and ΔvapHb but also ω of organic compounds, which further demonstrates the universality, stability, and accuracy of our proposed method.

A new universal method to predict the properties of organic compounds named as the positional distributive contribution method has recently been proposed by our group, which has successfully been used for the prediction of critical properties (Tc, Pc, Vc, Zc), normal boiling point (Tb), normal melting point (Tm), enthalpy of vaporization at the normal boiling point (ΔvapHb), and acentric factor (ω). In the following study, the positional distributive contribution method is further extended for the prediction of the flash-point (FP) temperature of various pure organic compounds. Comparison results between experimental and predicted data provide very satisfactory results. The overall average absolute difference (AAD) for the FP prediction of 287 organic compounds is 3.77 K, and the overall average relative deviation (ARD) is 1.16 %. However, Stefanis, Constantinou, and Panayiotou’s second-order group contribution (SCP-GC) method for the prediction of FP temperature of 287 organic compounds leads to the AAD and ARD of 14.16 K and 4.64 %, respectively. In addition, 342 organics missing experimental flash-point data are predicted via the presented procedure. More importantly, the good prediction capability of the proposed method shown in our previous works and this work suggests that it could be realized to use the same universal framework to predict not only Tc, Pc, Vc, Zc, Tb, Tm, ΔvapHb, and ω, but also the FP of organic compounds containing various functional groups, which further demonstrates the universality and stability of our proposed method.

Recently, our laboratory proposed a new universal method for the prediction of properties of organic compounds, such as the critical temperature, critical pressure, critical volume, critical compressibility factor, normal boiling point, and melting point. Here, the positional distributive contribution method is extended for the prediction of the enthalpy of vaporization of organic compounds at their normal boiling point (ΔvapHb). In this method, the position factor was used to take into account longer distance interactions, which could distinguish the overall isomer including the cis- and trans- or Z- and E- structure of organic compounds for their thermodynamics properties. The results indicate that our model provides very satisfactory results. The overall average absolute difference for ΔvapHb prediction of the 311 organic compounds is 1.00 kJ·mol−1, and 2.7 % is the relative deviation. Compared to those currently used prediction methods (including Riedel, Chen, Vetere, Liu, and Joback and Reid), the new proposed method could make significant improvements both in accuracy and in stability without requiring any input property. The most important point must be claimed is that all of those properties (Tc, Pc, Vc, Zc, Tb, and Tm) mentioned above and ΔvapHb in this work are predicted by the totally same universal positional distributive function and group framework proposed.

On the basis of the group contribution and position distribution function, a simple and accurate model to predict the critical compressibility factor, Zc, of organic compounds is presented in this study. The proposed model is developed to estimate Zc of a variety of pure organic compounds involving a carbon chain from C2 to C18. Comparison results between experimental and calculated data indicate that our model provides very satisfactory results. The overall average absolute errors for Zc predictions of 167 organic compounds is 0.007 with 2.45 % mean absolute relative derivation, which is compared to 0.018 and 6.90 % with the method of Joback and Reid, 0.016 and 5.94 % with the method of Constantinou and Gani, 0.012 and 4.73 % with the method of Wang et al., and 0.010 and 3.83 % with the method of Lee−Kesler. Also good prediction of the proposed method shown in our previous works and this work suggests that it is possible to use a similar framework to predict the critical properties, not only Tc, Pc, and Vc, but also Zc, of organic compounds containing various functional groups, which further demonstrates the universality of our proposed method.

In this study, a combined approach of group contribution and position distribution function is presented to estimate the critical volume of a variety of pure organic compounds involving a carbon chain from C2 to C18. The results indicate that our model provides very satisfactory results, and the deviation from the most reliable experimental data is not more than 2.1 %. The overall average absolute difference for critical volume Vc predictions of 219 organic compounds is 8.8 cm3·mol−1 and 2.1 % means relative derivation, which is compared to 26.5 cm3·mol−1 and 6.0 % with the method of Joback and 23.5 cm3·mol−1 and 5.6 % with the method of Constantinou and Gani. The higher prediction accuracy of the proposed method shown in our previous works and this work suggests that it is possible to use a similar framework to predict the three critical properties, Tc, Pc, and Vc of organic compounds containing various functionalities.

The search and development of anti-HIV drugs is currently one of the most urgent tasks of pharmacological studies. In this work, a quantitative structure–activity relationship (QSAR) model based on some new norm indexes, was obtained to a series of more than 150 HEPT derivatives (1-[(2-hydroxyethoxy)methyl]-6-(phenylthio)thymine) to find their pEC50 (the required effective concentration to achieve 50% protection of MT-4 cells against the cytopathic effect of virus) and pCC50 (the required cytotoxic concentration to reduce visibility of 50% mock infected cell) activities. The model efficiencies were then validated using the leave-one-out cross validation (LOO-CV) and y-randomization test. Results indicated that this new model was efficient and could provide satisfactory results for prediction of pEC50 and pCC50 with the higher Rtrain2 and the higher Rtest2. By using the leverage approach, the applicability domain of this model was further investigated and no response outlier was detected for HEPT derivatives involved in this work. Comparison results with reference methods demonstrated that this new method could result in significant improvements for predicting pEC50 and pCC50 of anti-HIV HEPT derivatives. Moreover, results shown in this present study suggested that these two absolutely different activities pEC50 and pCC50 of anti-HIV HEPT derivatives could be predicted well with a totally similar QSAR model, which indicated that this model might have the potential to be further utilized for other biological activities of HEPT derivatives.Based on atom characters and its position of the molecule structure, some norm indexes (including the norm(MD,1), the norm(MD,2) and the norm(MD,fro)) were proposed. And a QSAR model was developed for predicting pEC50 and pCC50 activities of anti-HIV HEPT derivatives. Results showed that these two activities pEC50 and pCC50 could be predicted well with a totally similar QSAR model. The leave-one-out cross validation and Y-randomization test results suggested the reliability and stability of this model, and this model might also be applied in a large response and structural domain by verified applicability domain.Download high-res image (88KB)Download full-size image

A new method is proposed based on the position group contribution additivity for the prediction of melting points of covalent compounds. The characteristics of this method are the use of position distribution function, which could distinguish between most isomers including cis or trans structure of organic compounds. Contributions for hydrocarbons and hydrocarbon derivatives containing oxygen, nitrogen, chlorine, bromine and sulfur, are given. Results are compared with those by the most commonly used estimating methods. The average derivation for prediction of normal melting temperature of 730 compounds is 14.46 K, compared to 29.33 K with the method of Joback, and 27.81 K with the method of Constantinou-Gani. The present method is not only more accurate, but also much simpler and more stable.

•LogEC50 of ILs on Vibrio fischeri is studied by topological method.•A general topological index was proposed.•A MLR model was developed to predict the toxicity of ionic liquids.As environmentally friendly solvents, ionic liquids (ILs) are unlikely to act as air contaminants or inhalation toxins resulting from their negligible vapor pressure and excellent thermal stability. However, they can be potential water contaminants because of their considerable solubility in water; therefore, a proper toxicological assessment of ILs is essential. The environmental fate of ILs is studied by quantitative structure–activity relationship (QSAR) method. A multiple linear regression (MLR) model is obtained by topological method using toxicity data of 157 ILs on Vibrio fischeri, which are composed of 74 cations and 22 anions. The topological index developed in our research group is used for predicting the V. fischeri toxicity for the first time. The MLR model is precise for estimating LogEC50 of ILs on V. fischeri with square of correlation coefficient (R2) = 0.908 and the average absolute error (AAE) = 0.278.