Figure 1 - uploaded by Helena Strömbergsson
Content may be subject to copyright.
![(a) The 3D structure of the enzyme cyclooxygenase in complex with the pain killer Diclofenac® (PDB code 1PXX [1]). (b) Details of Diclofenac®cyclooxyenase interactions.](profile/Helena-Stroembergsson/publication/265080879/figure/fig4/AS:668509776388099@1536396559636/a-The-3D-structure-of-the-enzyme-cyclooxygenase-in-complex-with-the-pain-killer.png)
(a) The 3D structure of the enzyme cyclooxygenase in complex with the pain killer Diclofenac® (PDB code 1PXX [1]). (b) Details of Diclofenac®cyclooxyenase interactions.
Source publication
Contexts in source publication
Context 1
... PCM modeling, descriptors of all dimensionalities have suc- cessfully been applied. Lists of 0D-3D descriptors are shown in Figure 10 together with a few examples of each descriptor category for the compound Diclofenac®. 0D descriptors are derived from the chemical formula, and 1D-descriptors are computed from a ligand represented as a substructure list. ...
Context 2
... to PCM is model induction and the main steps in this process are illustrated in Figure 11. Prior to model induction, raw data needs to be pre- processed to remove redundant entries and extreme outliers. ...
Context 3
... section contains an introduction to the supervised machine-learning process as well as the methods that have been used in PCM modeling. The supervised learning process usually begins with dataset partitioning followed by model induction, validation and interpretation ( Figure 11). ...
Context 4
... training set is used for model induction, the validation set can be used to select the best model, and the test set is used for assessment of the generalization error of the final model. It is important to select a representative training set that covers a large part of the data space as Figure 12 illustrates. For instance, a training set that only consists of ob- jects from square A in Figure 12 would result in a model that cannot predict objects from the remaining squares. ...
Context 5
... is important to select a representative training set that covers a large part of the data space as Figure 12 illustrates. For instance, a training set that only consists of ob- jects from square A in Figure 12 would result in a model that cannot predict objects from the remaining squares. Figure 12. ...
Context 6
... classification, the algorithm is designed to classify objects into one or several predefined discrete classes. Panel A in Figure 13 shows a training set with objects that belong to two classes (¨ and ). The classification func- tion, shown in panel B, provides the best separation of the objects, and can be used to classify new objects for which the class is unknown. ...
Context 7
... B Figure 13. Graphical representation of a classifier that maps objects in a training set (A) described by two variable (x, y) into two classes by a classification function (B). ...
Context 8
... regression model is a function that maps objects to a real-valued outcome variable. For example, panel A in Figure 14 shows a training set to which a regression function can be fitted. This is shown in panel B, along with a pre- diction of the real-valued outcome for a new unseen object. ...
Context 9
... x y y Figure 14. Graphical interpretation of a regression function (y=f(x)) induced from training set objects (A) and used to predict the outcome of a new unseen object (B). ...
Context 10
... classification, the performance of a model is often visualized by a so-called confusion matrix. Figure 15 shows a confu- sion matrix for a binary classifier with the two outcomes, high and low bind- ing affinity. The complexes that are correctly classified are denoted as true positives (TP) and true negatives (TN), and the complexes that are misclassi- fied are denoted as false positives (FP) and false negatives (FN). ...
Context 11
... area under the ROC curve close to 1.0 means that the classifier is able to perfectly map objects into classes, while an area close to 0.5 means that the classifier does not perform better than random guessing. A ROC curve of a moderately good classifier is shown in Figure 16. In regression analysis, there are several ways to quantify the extent to which the predicted outcomes differ from the actual outcome values. ...
