Antitumor Agents 252. Application of validated QSAR models to database mining: discovery of novel tylophorine derivatives as potential anticancer agents

Zhang, Shuxing; Wei, Lin-Hung; Bastow, Ken; Zheng, Weifan; Brossi, Arnold; Lee, Kuo Hsiung̀; Tropsha, Alexander

doi:10.1007/s10822-007-9102-6

Cited by 85 publications

(51 citation statements)

References 39 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…A widely used approach to establish the model robustness is so called Y-randomization (randomization of response) [23]. It consists of repeating the calculation procedure with randomized response values and subsequent probability assessment of the resultant statistics.…”

Section: Introductionmentioning

confidence: 99%

On Some Aspects of Variable Selection for Partial Least Squares Regression Models

2008

View full text Add to dashboard Cite

This paper tries to explore the optimum variable selection strategy for Partial Least Squares (PLS) regression using a model dataset of cytoprotection data. The compounds of the dataset were classified using K-means clustering technique applied on standardized descriptor matrix and ten combinations of training and test sets were generated based on the obtained clusters. For a particular training set, PLS models were developed with a number of components optimized by leave-one-out Q 2 and then the developed models were validated (externally) using the test set compounds. For each set, PLS model was initially constructed using all descriptors (variables). The variables having least standardized values of regression coefficients were deleted and the next model was developed with a reduced set of variables. These steps were performed several times until further reduction in number of variables did not improve Q 2 value. In each case, statistical parameters like predictive R 2 (R 2 pred ), squared correlation coefficient between observed and predicted values with (r 2 ) and without (r 2 0 ) intercept and Root Mean Square Error of Prediction (RMSEP) were calculated from the test set compounds. In case of all ten sets, Q 2 values steadily increase on deletion of variables while R 2 pred values do not show any specific trend. In no case, the highest Q 2 and highest R 2 pred appear in the same trial, i.e., with the same combinations of variables. This suggests that from the viewpoint of external predictability, choice of variables for PLS based on Q 2 value may not be optimum. Moreover, a clear separation of r 2 and r 0 2 curves in some sets suggests that such models may not be truly predictive in spite of acceptable R 2 pred values. Another observation is that coefficient of determination R 2 for the training set is more immune to changes on deletion of variables than the validation parameters like Q 2 and R 2 pred . Finally, a new parameter r m 2 has been suggested to indicate external predictability of QSAR models.

show abstract

Section: Introductionmentioning

confidence: 99%

On Some Aspects of Variable Selection for Partial Least Squares Regression Models

2008

View full text Add to dashboard Cite

show abstract

“…Some important factors affecting the anticancer activity of this kind of compound were selected out and then an optimal QSAR model showing a significant statistical quality and predictive ability was established. The predictive ability of the QSAR model was further confirmed by the small difference between the predicted and experimental cytotoxicity (pIC 50 ) for an external test set comprised of six recently reported congeneric compounds [17]. In this paper, based on obtained QSAR equation containing easily controlled parameters with simple and clear physical meanings, ten new compounds with higher anticancer activity have been theoretically designed.…”

Section: Introductionmentioning

confidence: 75%

“…Therefore, the QSAR model (Eq. 2) was further used to predict the activities of six recently synthesized congeneric compounds (see Table 4) as an external test set from the literature [17]. From Table 4, the predicted deviations lie in a range of À 0.71 -0.67, which is very near to the range (À 0.71 -0.58) of regression deviations for Eq.…”

Section: Qsar Equationmentioning

confidence: 87%

QSAR Studies and Molecular Design of Phenanthrene‐based Tylophorine Derivatives with Anticancer Activity

Liao

Chen

et al. 2008

QSAR Comb. Sci.

View full text Add to dashboard Cite

The Quantitative Structure -Activity Relationship (QSAR) of a series of novel phenanthrene-based tylophorine derivatives with anticancer activity has been studied by using the Density Functional Theory (DFT), Molecular Mechanics (MM2), and statistical methods. The established model shows not only significant statistical quality, but also predictive ability, with the square of adjusted correlation coefficient (R 2 A ¼ 0.813) and the square of the crossvalidation coefficient (q 2 ¼ 0.748). We find that the anticancer activity expressed as pIC 50 , which is defined as the negative value of the logarithm of necessary molar concentration of this series of compounds to cause 50% growth inhibition against the human A549 lung cancer cell line, closely relates with the energy (E HOMO ) of the Highest Occupied Molecular Orbital (HOMO), the net charge of the terminal H atom of substituent R 2 (Q HR2 ), the hydrophobic coefficient of substituent R 2 (log P R 2 ), and the net charges of the first atom of substituent R 1 (Q FR 1 ). The same model was further applied to predict the pIC 50 for six recently reported congeneric compounds as external test set, and the predicted pIC 50 values are close to the experimental ones, and thus it further confirms that this QSAR model has high predictive ability. The theoretical results can offer some useful references for understanding the action mechanism and designing new compounds with anticancer activity. Based on this QSAR equation, ten new compounds with higher anticancer activity have been theoretically designed and are expecting the experimental confirmation.

show abstract

“…The optimal 2D-QSAR model was built from the training set, and it was determined with the test set to confirm its predictive ability. A QSAR model is considered to own high predictive power only if its square of predictive correlation coefficient R 2 pred between the experimental and predicted activities is greater than 0.6 for the test set [36]. R 2 pred value is calculated by using…”

Section: Discussionmentioning

confidence: 99%

A Combined 2D‐ and 3D‐QSAR Study on Analogues of ARC‐111 with Antitumor Activity

Liao

et al. 2008

QSAR Comb. Sci.

View full text Add to dashboard Cite

Two-Dimensional (2D) and Three-Dimensional (3D) Quantitative Structure -Activity Relationships (QSARs) of a series of analogues of ARC-111 with antitumor activity expressed as pIC 50 , which is defined as the negative value of the logarithm of necessary molar concentration of these compounds to cause 50% growth inhibition against tumor cell lines P388, have been studied by using a combined method of the DFT, MM2, and statistics for 2D, as well as the Comparative Molecular Field Analysis (CoMFA) for 3D. The established 2D-QSAR model shows not only significant statistical quality, but also predictive ability, with the square of adjusted correlation coefficient R 2 A of 0.869 and the square of the crossvalidation coefficient q 2 of 0.834 as well as the square of predictive correlation coefficient R 2 pred of 0.867 for the test set. The 3D-QSAR model also shows good correlative and predictive capabilities in terms of R 2 (0.993) and q 2 (0.674) when CoMFA fields are used for the whole studied compounds. The results from 2D-and 3D-QSAR analyses conformably show that the steric interaction plays a crucial role in determining the cytotoxicity of the compounds and that selecting a moderate-size substituent R as well as increasing the negative charge of C 28 at the same time are advantageous to improving the antitumor activity. Such results might offer some useful theoretical references for understanding the action mechanism and directing the molecular design of this kind of compound with antitumor activity.

show abstract

Antitumor Agents 252. Application of validated QSAR models to database mining: discovery of novel tylophorine derivatives as potential anticancer agents

Cited by 85 publications

References 39 publications

On Some Aspects of Variable Selection for Partial Least Squares Regression Models

On Some Aspects of Variable Selection for Partial Least Squares Regression Models

QSAR Studies and Molecular Design of Phenanthrene‐based Tylophorine Derivatives with Anticancer Activity

A Combined 2D‐ and 3D‐QSAR Study on Analogues of ARC‐111 with Antitumor Activity

Contact Info

Product

Resources

About