FRED FAQ

Is it possible to access the internal file of probes in FRED_RECEPTOR? I would like to visualize the compounds docked into the defined active site, or any file using probes.
Unfortunately, the molecular shape probes are compiled into the program, and not accessible as a separate file. The probes are mostly lead-like molecules with an average of 22 heavy atoms. The probes were generated by clustering a library of lead-like molecules by shape. The resulting cluster centers are the probes that FRED uses.


Can I filter the FRED results according to the number of H-bonds made; for example, select only those which make 2 or more h-bonds to the receptor?
No, it is not possible to have the scoring function simply report the number of hydrogen bonds made. It is, however, possible to have FRED filter out molecules that do not make hydrogen bonding contacts with specific residues. You can do this by setting up hydrogen constraints in receptor file using the FRED_RECEPTOR program. Load the receptor into the FRED_RECEPTOR program, select constraints and then click on the protein atoms you want your fragments to make hydrogen bonding interactions with. After that just make sure the constraints are enabled (there is a flag in the receptor file that you can toggle with) and run your docking as normal.

During a recent docking run, I noticed that, in comparing the "*_scores.txt" file to the "*_alt_scores.txt" file, there were some lower scoring poses for some molecules in the latter file not present in the former. For the run, I set num_alt_poses=2.

For example, the first molecule in the *_scores.txt files reads:

Whereas the first 3 entries in the *_alt_scores.txt files are:

And you can see that pose 2 has a lower score than pose 1. This occurs for other molecules in the set as well. Am I interpreting the results incorrectly?

What's happening here is that FRED is choosing the best pose by a consensus of several scoring functions, not just the Chemgauss3 scoring function, even though the Chemgauss3 score is the only one being used to compare different ligands in your database.

If you find this behavior undesirable, you can disable the consensus pose selection by setting the following flags

-pose_select_weight_shapegauss 0
-pose_select_weight_plp 0
-pose_select_weight_chemgauss 0
-pose_select_weight_chemgauss2 0
-pose_select_weight_chemgauss3 0
-pose_select_weight_chemscore 0

I get the following when trying to run FRED: "ERROR: Unable to find an appropriate installation of fred for this architecture (osx-10.7-x86)". How can I get FRED to run?
FRED uses a script that utilizes uname to work out the current OS version. For some reason Apple revved the Darwin version to 10.7 causing this problem. (You can check this by typing 'uname -v' in a terminal.) The solution is to call the FRED executable directly like so:
/Applications/Openeye/arch/osx-10.6-g++4.2-x64/fred/2.2.5/bin/fred