Consensus prediction

Click Secondary Structure Consensus Prediction enabling the simultaneous execution of a number of selected prediction methods. The predicted secondary structures including the consensus secondary structure (Sec. Cons.) is returned (Figure 12.7). 

Figure 12,7. Secondary structure consensus prediction at NPS . Network Protein Sequence Analysis (NPS ) offers numerous methods for secondary structure prediction of proteins. The secondary structure consensus prediction (Sec.Cons.) of duck lysozyme is derived from simultaneous execution of predictions with more than one methods.

Maker. 3 The field is too young for a definitive standard, and the methods tend to perform better in the hands of experts, although this, perhaps, is changing. The recurring recommendation is that one develops a consensus prediction from several methods with several homologous sequences, if they are available. 

The second improvement in this area is the use of more involved consensus prediction methods. With a range of different methods at hand, it is possible to derive a consensus prediction as has been done in SOPMA... 

Geourjon, C. and G. Deleage, SOPMA significant improvements in protein secondary structure prediction by consensus prediction from multiple alignments. Comput Appl Biosci, 1995. 11(6) p. 681-4. 

Ijalsma, H. and van Dijl, J.M. (2005) Proteomics-based consensus prediction of protein retention in a bacterial membrane. Proteomics,... 

The accuracy of the consensus prediction regularities obtained by these three strategies is evaluated according to four indicators of the recognizing and predicting abilities of the integral decision rule, i.e., the results of self-prediction, leave-one-out cross-validation, split-half cross-validation, and double leave-one-out cross-... 

Here refers to the number of residues actually considered for every prediction, (cons JPRED consensus prediction dsc DSC mul MULPRED nnssp NNSSP orig phd PhD in its most current implementation phd PhD as run by JPRED pred PREDATOR psipred PSIPRED zpred ZPRED). Q3 refers to the three-state accuracy of a given prediction. 

Random Forest methods (Breiman 2001 Random Forests 2001) construct ensembles of trees based on multiple random selections of subsets of descriptors and bootstrapping of compounds. The compounds not selected in a particular bootstrapping are considered as a so-called out of bag set, and used as the test set. The trees are not pruned. Best trees in the forest are chosen for consensus prediction of external compounds. The method can include bagging (Berk 2008 Breiman 1996) and boosting (Berk 2008 Breiman 1998) approaches. 

Our previous experience suggests that the consensus prediction, which is the average of predicted activities over all predictive models, always provides the most stable results (Zhang et al. 2008 Zhu et al. 2008), and thus natarally avoids the need for (the best) model selection based on the statistics for the training and test sets. The consensus prediction of biological activity for an external compoimd on the basis of several QSAR models is more reliable and provides better justification for the experimental exploration of hits. 

We have used consensus prediction in many studies (de Cerqueira et al. 2006 Kovatcheva et al. 2005 Shen et al. 2004 Votano et al. 2004 Zhang et al. 2007, 2008 Zhu et al. 2008) and have shown that in most cases it gives better prediction and coverage than most of the individual predictive models. Thus, we recommend using consensus prediction for virtual screening of chemical databases and combinatorial libraries for finding new lead compounds for drug discovery. 

For the most part aU models succeeded in achieving reasonable accuracy of external prediction especially when using the AD. It then appeared natural to bring aU models together to explore the power of consensus prediction. Thus, the consensus model was constructed by averaging all available predicted values taking into account the applicability domain of each individual model. In this case, we could use only 9 of 15 models that had the AD defined. Since each model had its unique way of defining the AD, each external compound could be found within the AD... 

---