Activation prediction

Expert systems have also been devised for predicting biological activity. Predicting biological activity is discussed further in Chapter 38. 

Ideally, the results should be validated somehow. One of the best methods for doing this is to make predictions for compounds known to be active that were not included in the training set. It is also desirable to eliminate compounds that are statistical outliers in the training set. Unfortunately, some studies, such as drug activity prediction, may not have enough known active compounds to make this step feasible. In this case, the estimated error in prediction should be increased accordingly. 

The functionality available in MedChem Explorer is broken down into a list of available computational experiments, including activity prediction, align/ pharmacophore, overlay molecules, conformer generation, property calculation, and database access. Within each experiment, the Web system walks the user through a series of questions that must be answered sequentially. The task is then submitted to a remote server, where it is performed. The user can view the progress of the work in their Web browser at any time. Once complete, the results of the calculation are stored on the server. The user can then run subsequent experiments starting with those results. The Web interface includes links to help pages at every step of the process. 

Activity prediction is based on a list of models (i.e., QSAR models, pharmacophore models, etc.) that are maintained on the server. There is a second level of access so that only authorized users may be allowed to add or delete model entries. 

The property calculation experiment offers a list of 34 molecular properties, including thermodynamic, electrostatic, graph theory, geometric properties, and Lipinski properties. These properties are useful for traditional QSAR activity prediction. Some are computed with MOPAC others are displayed in the browser without units. A table of computed properties can be exported to a Microsoft Excel spreadsheet. 

Figure 20 Feed-forward neural network training and testing results with back-propagation training for solvent activity predictions in polar binaries (with learning parameter rj = O.l).

A Brief Review of the QSAR Technique. Most of the 2D QSAR methods employ graph theoretic indices to characterize molecular structures, which have been extensively studied by Radic, Kier, and Hall [see 23]. Although these structural indices represent different aspects of the molecular structures, their physicochemical meaning is unclear. The successful applications of these topological indices combined with MLR analysis have been summarized recently. Similarly, the ADAPT system employs topological indices as well as other structural parameters (e.g., steric and quantum mechanical parameters) coupled with MLR method for QSAR analysis [24]. It has been extensively applied to QSAR/QSPR studies in analytical chemistry, toxicity analysis, and other biological activity prediction. On the other hand, parameters derived from various experiments through chemometric methods have also been used in the study of peptide QSAR, where partial least-squares (PLS) analysis has been employed [25]. 

Even as the computational prediction error rate is reduced to acceptable levels, many cases will be encountered in which the predictions are indistinguishable to within error. In a scenario in which several different in silico designs are given equivalent but favorable activity predictions, the end user s medicinal experience may help decide which to promote to synthesis. The quality of that decision at this point will be strongly influenced by how easy it is to understand the different contributions to the computational predictions. Interpretability is thus critical for synergistically utilizing the experience of the end user. 

Petrii OA, Tsirlina GA. 1994. Electrocatalytic activity prediction for hydrogen electrode reaction intuition, art, sceince. Electrochim Acta 39 1739-1747. 

Ozdemir V, Posner P, Collins EJ, Walker SE, Roy R, Walkes W et al. CYP1A2 activity predicts clozapine steady state concentration in schizophrenic patients [abstract], Clin Pharmacol Therapeutics 1999a 65 175. 

Low DPD activity predicts decreased response and increased toxicity... 

Tab. 5.3 Correlation coefficients between the similarity-based and hypothesis-based activity prediction scores and experimental activity values, taken over the whole Cox2 inhibitor set and with...

Fig. 30. KINPTR activity predictions, adiabatic pilot plant data. 

Predict the effect on oxidation of ketone bodies and of glucose in nonhepatic tissue of individuals with markedly diminished /3-oxyacid-CoA-transferase activity. Predict the effect if the activity was absent. 

The present article is not designed to review the work devoted to theoretical treatment of multiple steady states and oscillatory activity predicted from these equations, for those readers who seek a profound review the texts by Aris (7), Schmitz (5), and others (7,9) are recommended. 

D. Haring and P. Schreier, Chemical engineering of enzymes altered catalytic activity, predictable selectivity and exceptional stability of the semisynthetic peroxidase seleno-subtilisin, Naturwissenschafien 1999, 86, 307-312. 

Using this activity, predict the ideal temperature for electroplating a piece of jewelry with copper. 

If new scaffolds are to be found, a second important feature of such models is their completeness. It allows different structural solutions to fulfil the interaction pattern required by the pharmacophore. Accuracy in the activity prediction, on the other hand, is not paramount. 

When compared with the first application (virtual screening), pharmacophore models for activity prediction require more sophisticated models to capture more subtle effects (e.g. orientation of directional features, different weights to... 

Regarding the statistics reporting the SAR data, we were rather lenient since accurate activity prediction was not our objective. Hence, models exhibiting a... 

In addition to this pharmacophore hypothesis, although it met only three of the four criteria, model 1 from run 6 was retained. Surprisingly, despite criterion number 2 not being satisfied (RMS= 1.62, r=0.79), this model exhibits a remarkable ability to discriminate between active and inactive compounds as assessed by the ROC curve, AUC=0.95. In contrast, model 1 from run 8 has good statistics (RMS=0.76, r= 0.96) but a lower AUC of 0.87. This illustrates that a good model for activity prediction may not be the best for virtual screening applications. Let us analyze these two pharmacophore hypotheses further. 

Although rather similar in their composition, hypotheses 1 and 2 would probably show different performances depending on their use. The preliminary statistics are clearly in favor of hypothesis 2 (see Fig. 15.11, top), making this model a good candidate for activity prediction. A more thorough validation would be required if it were to be used for this purpose. In particular, further assessment with compounds external to the training set would be necessary. 

---