StoneHinge is a hinge prediction algorithm that peforms a network-based analysis of a single protein structure to detect hinge regions. StoneHinge incorporates both ProFlex, which uses the FIRST constraint counting algorithm, and DomDecomp, which uses Gaussian Normal Mode analysis.
For the StoneHingeP (ProFlex-based) predictions, StoneHinge finds the two largest rigid domains in the protein at each bond coordination level. The "native" state is taken to be the coordination level with the biggest second-largest rigid domain. The hinges are then taken to be any residues between these two domains.
For the StoneHingeD (DomDecomp-based) predictions, StoneHinge identifies hinges as residues in between domains or at domain boundaries. The StoneHingeP and StoneHingeD predictions are then combined by taking any StoneHingeD hinge within five residues of a StoneHingeP hinge.
To submit a protein, go to the submission page. After submission, processing will take several minutes, and you will then be brought to the viewer page. This page allows you to view and rotate the protein in a JMol applet. If you cannot see this, make sure that Java is installed on your system. The links to the right allow you to highlight the predicted domains and hinges, and the hinge residues are listed at the bottom of the page.
The bottom of the page also includes two links. The first of these is to the detailed StoneHinge output. This will list any errors that occured, as well as give more information about the StoneHingeP and StoneHingeD predictions.
The StoneHingeP output first lists information about the state chosen during the simulated denaturation:
Next, the size and residues involved in the two largest rigid clusters are listed.
StoneHingeP then lists the predicted hinges. The range of residues is given for each hinge, along with its predicted flexibility. Any flexibilities less than zero are predicted to be flexible regions of the protein. The larger the magnitude of the flexibility, the more flexible the hinge is predicted to be.
Each hinge is also broken down into coordinated flexible residues. These residues are part of the hinge and predicted to move in unison. The flexibilities for these coordinated residues are given, using the same scale as hinge flexibility.
Residues in StoneHingeD domains and hinges are listed.
The viewer page also includes a link to the hydrogen-bond dilution plot, as generated by ProFlex. If this link is not present, it means that ProFlex was not able to succesfully generate a plot for this protein. This plot shows rigid regions in the protein over the entire course of the simulated denaturation. This first line corresponds to the start of the denaturation when the simulated temperature is very low and all of the potential hydrogen bonds are included. Each subsequent line corresponds to a more flexible state of the protein resulting from an increase in the simulated temperature. Each line of the plot contains information about all residues, with the amine terminus at the left end and the carboxyl terminus at the right end. Each rigid residue is colored according to which rigid cluster it belongs to; all residues of the same color belong to the same cluster. Flexible residues are shown as a horizontal line.
Before the prediction is made, StoneHinge removes all heteroatoms from the structures. This includes inhibitors, ligands, and cofactors, as these stabilize the protein in the current conformation which causes the hinge to no longer appear flexible. Hydrogen atoms are added to the structure using Gromacs, as they are required for ProFlex to calculate the hydrogen bond energy. In addition, all water molecules are removed from the structure. (Using the command-line version of StoneHinge, it is possible to include internal water molecules. These can improve the hydrogen bond dilution profile, but typically have little effect on the hinges predicted.)
If you have any additional questions about StoneHinge, e-mail stonehinge@gmail.com.
Reference: Keating KS, Flores SC, Gerstein MB, Kuhn LA (2009). StoneHinge: Hinge Prediction by Network Analysis of Individual Protein Structures. Protein Science 18: 359-371.