Fragment screening allows you to screen a database of small chemical fragments to a protein receptor. The fragments are then ranked by p-value providing a quantitative measures of confidence that a fragment is a true active and binds in the specific pose in absolute terms, instead of just ranking it versus other fragments. The fragment hits can then be "grown" or linked to produce a lead with high affinity. In this example we screen a database of fragments to cAMP-dependent protein kinase ATP binding pocket.

Step 1 For this example we will use PDB file 1REK. It needs to be converted to an ICM Object and the waters and pentanal should be deleted.

Step 2 Move the ligand out of the object into a separate object so it is not included in the docking maps. Select the binding site by right clicking on the ligand and choose "Select Neighbors" in this example a radius of 3.5A is sufficient to select the pocket. Be careful not to select residues outside of the pocket because this could potentially include smaller cavities outside the pocket where fragments could bind.

Step 3 Setup the docking project using the docking menu and the option "Setup Receptor".

Step 4 The probe represents the docking starting position. Generally you can keep it in the default position. If you want to move it use the middle mouse button and hold the SHIFT button for global rotation. .

Step 5 The purple box represents the region in which maps will be generated. To change the box size you can use the left mouse button with the cursor at any corner of the purple box to change it. It is important for fragment docking to minimize the number of cavities outside the pocket so to prevent fragments from docking outside of the pocket. Make sure the purple box is as close to the pocket as possible. .

Step 6 To start the screening go to Docking/Scan Probes Fragments. A database of fragments is provided with the database distribution - the file is called fragments.icf.

Step 7 The screen will take several minutes to run in the background.

Step 8 Once the screening has finished a table of the results will be displayed. The docked fragments can be viewed by toggling the view on and off using the buttons in the "L" column. Guide to the table columns: IX is the index number from the docked database Score is the ICM score -32 and lower are generally considered good scores - but depends on the receptor (e.g. exposed pockets or pockets with metal ions may have higher scores than -32). Natom is the number of atoms in docked ligand Nflex is the number of rotatable torsions. Hbond is Hydrogen Bond energy Hphob is the hydrophobic energy in exposing a surface to water VwInt is the van der Waals interaction energy (sum of gc and gh van der waals). Current version of the score uses explicit van der Waals interaction energy calculation (no grids) Eintl is internal conformation energy of the ligand Dsolv is the desolvation of exposed h-bond donors and acceptors. SolEl is the solvation electrostatics energy change upon binding. mfScore is the potential of mean force score pValue and Percentile. Because the fragment database is relatively small it has been possible to pre-validate the scoring individually for each member of the library. This allows us to estimate sensitivity and specificity (true positive/true negative rates) at any given score cutoff. As a result, expected number of false positive binding poses below the cutoff for each particular fragment can be generated. This provides a quantitative measures of confidence that a fragment is a true active and binds in the specific pose in absolute terms, instead of just ranking it versus other fragments (see Step 9). The p-value represents the probability of obtaining a score this good or better by random chance — so lower p-values are better. Lower percentile indicates better score (e.g. being among top 1% is better than among top 5%). examplePDB Each fragment in the library is linked back to the PDB entries with complexes of the fragment-containing ligands, and appropriate rotation/translation matrices for instant superposition. Click on the hyperlink to display the fragments. Location these values indicate the location of the ligand in the pocket. The values can be clustered to select fragments in different regions of the pocket (Step 10-12).

Step 9 Sort the table by pvalue to rank the fragments.

Step 10 To cluster the fragments by location in the pocket. Select the .Location. column and click on the cluster button.

Step 11 Select representative centers from the cluster tree by lowest p-value. Selections will be highlighted in blue in the tree and the table.

Step 12 Toggle through the selected fragments to choose lead candidates for linking using the Ligand Editor.

Reload Fragment Screening Results To reload the fragment screening results:

• Docking/Open Project and locate the directory where your fragment screening results are located.
• Select Docking/Make Hitlist and browse for the Docking_Project_Name_Fragments.ob file which contains the results of the screen.
• A table as shown in Step 8 will be displayed.