Contents
Important Tips: There might be an error 'Application Blocked by Security Settings' when you start the 3d viwer (AstexViewer) in your browser. Please set the security levels to 'Medium' through the following steps.
1. Go to the Start menu, click 'Configure Java'. 2. In the Java Control Panel, click on the Security tab. 3. Select the Security level 'Medium'. 4. Click Apply. 5. Click OK to save changes made to the Java Control Panel.
Introduction
PredHS is a flexible, interactive webserver to predict hot spots on protein-binding interfaces by using three main sources of information, namely site, Euclidean, and Voronoi features describing the properties of either the target residue or the target residue’s structural neighborhood. It integrates a set of 38 optimal features selected from 324 site, Euclidean, and Voronoi properties by a twostep feature selection method. Figure 1 shows the framework of PredHS [1].
Figure 1. The framework of PredHS. (A) Feature representation: We encode each interface residue using 108 site features, 108 Euclidean neighborhood features, and 108 Voronoi neighborhood features. (B) Two-step feature selection: the first step of feature selection is done by a random forest algorithm, we select the top 77 features with Z-Score larger than 2.5; the second step is performed using a wrapper-based feature selection. Features are evaluated by 10-fold cross-validation with the SVM (support vector machine) algorithm, redundant features are removed by sequential backward elimination. (C) Prediction models: PredHS-SVM and PredHS-Ensemble. For PredHS-Ensemble, an ensemble of n classifiers is built using different subsets, the final result is determined by majority votes among the outputs of the n classifiers.
Getting Started
Input data can be structure files in PDB format or PDB codes. The input structure should contain at least two chain identifiers forming the interface and the interface definition. One can query multiple structures in one run (maximum allowed 10 structures). Please follow the descriptions for the input format. Users could leave their email address; PredHS will send the prediction results to the address. Titles also could be specified for the user to distinguish their different jobs. Private Key is set to protect your structure and analysis. Please refer to Figure 2.
Figure 2. Job submission.
The server will check the validity of the input structure, and once confirmed, process to the secend step to select the query protein and its partners. If the selection is done, please click the button "submit" to run the job (Figure 3). Users will then be directed to the result page with job status. Book the link if you want to check your results later. However, the analysis could also be retrieved by user email or job ID in the result page.
Figure 3. Select the query protein and its partners.
PredHS will return two lists of residue ids which were predicted to be hot spots corresponding to the SVM method and the Ensemble method respectively. The red residues in the query sequence are predicted hot spots. Users can download the results in text. By default, PredHS predicts a interface residue to be in a hot spot when the associated score is higher than 0 but this cutoff is adjustable by the user. In order to visualize the 3D structures of the prediction resutls, users can click "View in 3D". (Figure 4)
Figure 4. Prediction results.
The information of the query job is shown at the top. The sturctures of the query protein and its partners will be shown in AstexViewer. Predicted hot spots are rendered in different colors from white to red, according to the probability from the lowest to the highest. Users can view the predicted hot spots with SVM output scores listed at the bottom, and move the mouse over the individual residues to display them in the molecular viewer.
Figure 5. Prediction results and visulization.
An example Please follow the link to see an example.
References [1]. Deng L, Zhang QC, Chen Z, Meng Y, Guan J and Zhou S. PredHS: a web server for predicting protein–protein interaction hot spots by using structural neighborhood properties. Nucleic Acids Research, 42(W1):W290-W295 (2014) [PubMed] [2]. Deng L, Guan J, Wei X, Yi Y, Zhang QC and Zhou S. Boosting Prediction Performance of Protein–Protein Interaction Hot Spots by Using Structural Neighborhood Properties. Journal of Computational Biology, 20(11):878–891 (2013) (Presented at RECOMB 2013)[PubMed]
