Regression-based QSAR

Eriksson, L., Jaworska, Worth, A.P., Cronin, M.T.D. and McDowell, R.M. (2003). Methods for reliability and uncertainty assessment and applicability evaluations of classification- and regression-based QSARs. Environmental Health Perspective 111 1361-1375. 

The best models to predict aquatic toxicity are ones that are simple and interpretable. A regression-based QSAR established with fundamental descriptors maximizes the interpretability of the model, while at the same time maintaining simplicity. Such QSARs are easily updated, capable of mechanistic-based interpretation, portable from one user to another, and allow the user to observe and comprehend how the prediction of toxic potency is made (Schultz and Cronin, 2003). 

QSARs utilized by the U.S. EPA for the prediction of the dermal uptake (absorption through the skin) of compounds are well described by Walker et al. (2002). Predictions of the ability of chemicals to be absorbed across the skin allowed for the potential of dermal toxicity to be assessed. Typically simple regression-based QSARs, which were based either on hydrophobicity and molecular size, or hydrophobicity alone, were utilised. 

In this step, one or more independent experts should evaluate the quality of the training set data along with any other available data for the endpoint predicted by the QSAR. This should enable an evaluation to be made of the maximal predictive capacity that could be expected for the QSAR. For QSARs, the inevitable variability in descriptor and response variable data should be taken into consideration when defining criteria for predictive capacity. For example, in the case of a regression-based QSAR, it might be decided that its predictions should fall within a specified prediction interval, and that the R2 value for predictions of independent data should exceed a specified value. Issues relating to the quality of data for use in QSARs are discussed in Cronin and Schultz (2003) and Schultz and Cronin (2003). 

Quantitative Assessment of Toxidty Mechanism of Action. . Regression-based QSAR models Receptor-based ligand screening... 

The limitations of QSAR for enzymes are related to the fact that the experimental measurement of kinetic parameters is inherently prone to errors. Kinetic constants for the same compound vary substantially among studies, depending on the enzyme source (recombinant enzyme, purified enzyme, subcellular fraction, etc.) or experimental conditions. Reported Vmax values for the same compound can vary by 2 to 3 orders of magnitude, seriously impacting regression-based QSAR modeling. Therefore much larger, consistent datasets for each enzyme will be required to increase the predictive scope of such models. 

Viswanadhan VN, Mueller GA, Basak SC, Weinstein IN. Comparison of a neural net-based QSAR algorithm (PCANN) with Hologram- and multiple linear regression-based QSAR approaches application to 1,4-dihydropyridine-based calcium channel antagonists. I Chem Inf Comput Sci 2001 41 505-11. 

Description of the type of model. This section specifies the type of model (e.g., SAR, regression-based QSAR, expert system, battery of (Q)SARs) and defines the endpoint and the dependent variable being modeled, reporting also information (if available) on the quality of the data used... 

Sagrado, S. and Cronin, M.T.D. (2006) Diagnostic tools to determine the quality of transparent regression-based QSARs the modelling power plot. J. Chem. Inf. Model, 46, 1523-1532. 

---