CHIME is a molecular graphics plugin that allows an Internet Browser to display PDB files. CHIME is freely available for download. Some computers (such as those running UNIX) are not currently able to use CHIME. Q-SiteFinder results can be downloaded and viewed in Rasmol. Instructions are given below.

The following information refers to the figure above.

1. The protein is displayed using the CHIME plug-in. Initially, the first predicted site (turquoise) and any ligands (yellow) are displayed. To rotate the molecule, click and drag with the left mouse button. To zoom in or out, hold down the shift key and click and drag with the left mouse button. To rotate the molecule in a plane, hold down the shift key and click and drag with the right mouse button. To translate the molecule, hold down the control (ctrl) key and click and drag with the right mouse button.

2. Predicted binding site selection is colour-coded according to the likelihood of being an actual binding site. Green is the most likely, followed by blue, purple and orange/brown.

3. This is the main option window. All of the predicted sites can be selected or deselected. CPK colours can be toggled on and off. Ligands can be hidden. The sites can be shown in the probe-cluster form, or all the residues close to the probe sites can be displayed. The PDB file can be downloaded (instructions for manipulation using Rasmol are given below).

4. This box displays the coordinates that define the binding-box around selected sites. This can be used as input for a docking program to ensure that ligands are only docked onto into the binding site. In the example shown, the binding site consists of predicted sites 1, 2 and 8.

5. This box gives information about the selected predicted binding site. It gives an estimate of cavity volume, protein volume and also shows the binding precision of known ligands.

6. All of the protein atoms close to a probe-cluster defining a site are shown here. The atom number, atom type, residue name, chain identifier and residue number are given. Each entry is clickable, and causes the selected atom to be displayed in CPK colours. They are invisible if the predicted sites are being displayed in CPK colours. These selected atoms are persistent, and are not affected by selection or deselection of sites. They are only deselected by clicking on 'hide atoms'.

Mage is a Java-based molecule viewer developed by the Richardson lab. To use Mage, you will need the Java 2 runtime environment. This is freely available from http://java.sun.com. The following information refers to the figure above.

1. The protein is displayed using Mage. Initially, the first predicted site (turquoise) and any ligands (yellow) are displayed. To rotate the molecule, click and drag with the left mouse button. Predicted binding site selection is colour-coded according to the likelihood of being an actual binding site. Green is the most likely, followed by blue, purple and orange/brown.

2. To zoom in or out, drag the button left or right. To display predicted sites as either favourably binding methyl probes, or residues close to the probes, click on the appropriate buttons.

3. This is the information selection window. Note that unlike the CHIME interface, predicted site information must be selected separately from site selection in Mage. The PDB file can be downloaded (instructions for manipulation using Rasmol are given below).

4. This box displays site information, including precision, estimated site volume, estimated protein volume and the coordinates that define the binding-box around selected sites. Binding box coordinates can be used as input for a docking program to ensure that ligands are only docked onto into the binding site.

5. All of the protein atoms close to a probe-cluster defining a site are shown here. The atom number, atom type, residue name, chain identifier and residue number are given. These entries are not clickable, unlike for the CHIME interface.

This data was obtained using 134 PDB entries that correspond to the GOLD data set. We defined success as having a precision of at least 25%. Precision is defined as the percentage of probe sites defined by a single cluster that are within 1.6 angstoms of an atom of a particular ligand. It is a measurement of how well the predicted site maps onto the ligand coordinates. When a single ligand gives a success in two separate sites, only the higher ranking site has been counted, since they are part of the same binding site. If two ligands are found in the same predicted binding site, only one success is counted. Separate ligands in separate sites are counted as successes separately.