KOWWIN program

After dissecting out the amino, amide, urea and -N=C fragments (all considered as aliphatic attached), the KOWWIN program seeks out structural corrections to allow for the fact that part of the H-bond acceptor strength ( f ) is lost when they are attached together. As Figure 5.1 shows, the program appears to overcorrect in this instance. 

As previously noted, the KOWWIN program does not consider any interaction of gemi-nal chlorines with each other. It also neglects any of their interaction with the nearby H-bond acceptors. Thus, as Table 5.7 shows, it starts out with a much higher value for each chlorine (0.3102 vs. 0.18 in ACD and 0.06 in CLOGP), but it then underestimates the total when three chlorines are on one carbon. As a result KOWWIN underpredicts dipterex by 0.79 log units and trichloroacetonitrile, C13CCN, by 0.88 log units (Measured = 2.09 KOWWIN = 1.21). 

The KOWWIN program calculates the -ONE and -OL forms very nearly the same and close to the measured value for both the thiopurine and thioguanosine. The almost unlimited freedom to incorporate atom strings to fit measured values certainly makes for improved statistics, but it might possibly be misleading when the measurement that is fit is actually for an equilibrium mixture of structures having quite different hydrophobic properties. 

A knowledge of physicochemical properties underpins many of the predictions of toxicity made by the EPA. In addition, properties are required for the estimation of fate. The logarithm of the octanol-water partition coefficient (log Kow) is one of the values for which predictions are most widely accepted (many agencies worldwide will accept predicted values for log Kow). The KOWWIN program as well as the other programs in the EPISUITE software noted in Table 19.2 are used routinely by the EPA to calculate the physicochemical properties of new chemical substances for PMNs. 

Quantitative structure-activity/pharmacokinetic relationships (QSAR/ QSPKR) for a series of synthesized DHPs and pyridines as Pgp (type I (100) II (101)) inhibitors was generated by 3D molecular modelling using SYBYL and KowWin programs. A multivariate statistical technique, partial least square (PLS) regression, was applied to derive a QSAR model for Pgp inhibition and QSPKR models. Cross-validation using the leave-one-out method was performed to evaluate the predictive performance of models. For Pgp reversal, the model obtained by PLS could account for most of the variation in Pgp inhibition (R2 = 0.76) with fair predictive performance (Q2 = 0.62). Nine structurally related 1,4-DHPs drugs were used for QSPKR analysis. The models could explain the majority of the variation in clearance (R2 = 0.90), and cross-validation confirmed the prediction ability (Q2 = 0.69) [ 129]. 

SRC. 1995b. Syracuse Research Corporation. KOWWIN. Log Kow program for MS Windows 3.1. Syracuse Research Corporation, Merrill Lane Syracuse, NY. 

SRC. 1995. Syracuse Research Center. Octanol-Water Partition Coefficient Program (KOWWIN, version 137, serial L0148). Chemical Hazard Assessment Divison, Environmental Chemistry Center, Syracuse, NY. 

Of the many listed by Mannhold, three programs, differing significantly in methodology, were chosen for comparison of performance KOWWIN, ACD/LogP, and CLOGP. As Table... 

Assignment of probable error to the calculation of log Poctis very difficult. Some of the programs (e.g., the three mentioned in the previous paragraph) retrieve a measured value when available, and this often is the most reliable measure of the error to be expected in solutes of similar structural complexity. KOWWIN offers no error estimate directly with... 

A computer calculation often can detect errors in published values that may not be immediately apparent to the average researcher. For instance, a measured log P(oct) value for toluene diamine was reported as 3.16 (Veith, 1979) but three computer programs show it to be three log units lower ACD = -0.36 KOWWIN = 0.16 CLOGP = 0.14. One then wonders if a typographical error occurred, and the measured value was 0.16, not 3.16. 

The user should see estimation as an experiment. Just as a good experiment is planned, performed, and accurately repeated, so should estimations be. There is a danger in computer-based methods that non-experts or even non-chemists can use. Because the programs can give decent answers without any fine-tuning or validation, it does not mean that best practice can be abandoned. The best experiments are reported with a confidence interval for a dependent variable, with hopefully much smaller uncertainties concerning the independent variable. This is harder to achieve when we are performing an estimation. Consider an example that will be returned to later the estimation of octanol-water partition coefficients (Kow). Take a method such as the SRC software KOWWIN to... 

The SRC program KOWWIN uses an atom/fragment contribution method to predict log Kow. This is a reductionist method (the fragment coefficients were derived by multiple regression from a development set of reliably measured log Kow values). The other main software tool, ClogP for Windows (Leo, 1993) is a constructionist method (the fragment coefficients are evaluated from the simplest examples in which they occur). Both methods have a high level of accuracy and are widely accepted as the best tools available. 

In terms of the descriptors used in the prediction of skin permeability coefficients, the octanol-water partition coefficient F is a well-established measure of hydro-phobicity (Dearden, 1990). There is a variety of algorithms to calculate log P, including Web-based programs and the KOWWIN software, which is part of the EPISuite utility. EPISuite is available free from the U.S. Environmental Protection Agency (see Cronin and Livingstone, 2004b, for more details EPISuite can be downloaded from http //www.epa.gov/oppt/exposure/docs/episuite.htm). Molecular size is well modeled by molecular weight, which of course is fimdamental and trivial to calculate. 

The descriptors were calculated here, for each complex, using the computer programs MOPAC/PM6, PRoperty Evaluation by CLAss Variables (PRECLAV) (Tarko 2005), and DESCRIPT (Tarko 2008a). The LogP descriptor was calculated using the KowWin algorithm of EPISuite software (EPISuite Meylan and Howard 1995). The type of chemical bonds was defined according to Table 4.1. 

Calculated using the KowWin soflware program located at www.esc.syrres.com (2003). 

---