Significance QSAR equations

Fischer statistics (F) Fischer statistics (F) is the ratio between explained and unexplained variance for a given number of degree of freedom. The larger the F value the greater the probability that the QSAR equation is significant. The F values obtained for these QSAR models are from 17.622 to 283.714, which are statistically significant at the 95% level. 

Examining the QSAR for P. putida TVA8 for the chemicals tested (which represent different classes of compounds), a highly predictive QSAR equation (Equation 17.1, R2= 0.90) was obtained that described induction by hydrophobicity alone. The ten chemicals tested all significantly induced the tod operon. This increase in induction correlated with the log Kow values. Other aromatic compounds, such as naphthalene, failed to induce TVA8. 

This sensor responds to octane. This compound is significantly more volatile than the other test componds, proving to be problematic in the development of a significant QSAR. The compound 3-methylheptane clearly does not fall into the same group as the n-alkanes. The value predicted for induction (111%) using Equation 17.5 is different from the actual value of 36%. Further testing with a wider range of branched alkanes would be necessary to determine if, like -alkanes, they formed a QSAR that was dependent on molar volume. 

Short Neurotoxins. Figure 1 shovs the fragments that were selected By the program for the QSAR equation. Figure 3 shovs the primary structure of some of the snake venoms. In these, ve have underlined the fragments that vere selected by the QSAR equation as significant to activity. 

It was also shown that the "ip index could be used as a significant second variable in the QSAR equation. Hansch-type analysis yielded r = 0.90. ... 

Significant progress in QSAR resulted from Hansch analyses of enzyme inhibitors [432, 456, 668 — 670], especially from the systematic work of Hansch and his group on dihydrofolate reductase and on cysteine and serine proteases. Most of our current knowledge of the quantitative aspects of ligand-protein interactions has been derived from QSAR equations, aided by the interpretation of the 3D structures of enzymes and their inhibitor complexes with molecular graphics [38, 288, 671 — 676]. 

Another extensively investigated field in QSAR are mutagenic agents [325 — 328, 757 — 760]. The QSAR equations of a series of l-(X-phenyl)-3,3-dialkyltriazenes show that mutagenic activity (and presumably carcinogenicity) can be minimized with relatively little loss in antitumor potency. While such hints are useful, they should not be overemphasized lipophilicity optima can be significantly different in isolated cells and in whole animals. 

With the data from TABLE 3, several QSAR equations were calculated for the whole data set based on one or more descriptors (TABLES). QSAR equations are given, based solely on hydrophobicity (equation 4), solely on descriptors for electronic effects (equations 5 to 7) and on a combination of these two descriptors (equations 8 to 10). All correlations, except in equations 5 and 6, are significant at P < 0.05 or better. 

One of the most frequently used physico-chemical descriptors of similarity is the logarithm of the octanol/water partition coefficient (log P, or log Kq, ) and its significance and application is well documented (Hansch and Leo 1979). While it has been recognized that any reliable prediction can only be made within a congeneric series of compounds (Rekker 1985), the a priori definition of congenericity is still problematic in many cases. Consequently, many QSAR equations are limited to very narrowly defined sets of substances and cannot be generalized. 

From a theoretical point of view, the proper application of regression analysis requires the formulation of a working hypothesis, the design of experiments (i.e., compounds to be tested), the selection of a mathematical model, and the test of statistical significance of the obtained result. In QSAR studies, this is pure theory. Reality is different QSAR studies are most often retrospective studies and in several cases many different variables are tested to find out whether some of them, alone or in combination, are able to describe the data. In principle, there are no objections against this method because QSAR equations should be used to derive new hypotheses and to design new experiments, based on these hypotheses. Then the requirements for the application of statistical methods are fulfilled. 

As for the modeling of in vitro and in vivo potencies and selectivities, some interesting models were obtained on a subset of representative ligands. In fact, the QSAR analysis of functional activities for this set of compounds, although numerically limited, yields successful correlations [97]. Among these linear correlations, the following equation constitutes a significant example ... 

---