FRED 2.2.1

New Features

  • Added support for OEB rotor offset compression when writing the MASC tagged version of the input database (provided the initial input database used rotor offset compression).

  • Chemscore’s, OEChemscores’s and Screenscores’s hydrogen bonding terms now only allow one hydrogen bond per hydrogen. Also prevented two hydroxyls from making both and acceptor-donor interaction and a donor-acceptor interaction.


  • Modified OEChemscore’s hydrogen bonding term to be more forgiving of non-ideal geometries. The range of geometries considered ideal is unchanged.

  • Corrected a deficiency in Chemgauss3 metal term and metal constraints. The metal chelator interaction function was picking up on some but not all of the allowable geometries for metal-chelator interactions.

  • Improved Chemscore’s, OEChemscore’s and Screenscore’s handling of rotatable hydrogens involved in hydrogen bonding by replacing the brute force torsion driving search for the optimal hydrogen position with an analytic solution for the best position.

  • Added a new flag “-no_masc_data_calc” which will prevent Fred from calculating MASC data for ligands. Any ligands that missing needed masc data will not be docked. The purpose of this flag is to prevent it from doing a lengthy MASC calculation when all but a handful of ligands have the required MASC data.

  • Extended initial list of atom types know to Chemgauss3. This should improve the startup speed of runs using Chemgauss3, by avoiding costly grid recalculations each time a new atom type is encountered. This change is only for speed, there is no change to the Chemgauss3 score value Fred calculates.

  • When using constraints fred now defaults to effectively using a -clash_scale value of 0.6 if the -clash_scale flag has not been set by the user. This helps eliminate unusually close protein-ligand contacts that can occasionally occur when constraints are used. The original behavior can be obtained by explicitly setting -clash_scale to 0. Runs without constraints are unaffected and behave as before.

  • Chemgauss2 is no longer used by default in the consensus pose stage of docking. This has no real effect on docking performance and will reduces Fred’s startup time.

Bug Fixes

  • Important: Fixed a bug when using -pro and -box to create a receptor that causes the inner contour to be set to an extremely small value. The resulting receptor would produce extremely poor docking results without warning. In this release newly created receptors will now have reasonable inner contours. Additionally Fred now checks if the inner contour volume is extremely small (i.e., one that was produced by this bug), and if one is detected the inner contour is turned off and a warning is issued before proceeding with the run.

  • Fixed a when requesting alternate poses during a PVM run, which caused the run to shutdown and not dock any molecules.

  • Fixed a bug in screenscore when the receptor/protein it was initialized with did not have explicit hydrogens. The bug causes the initial setup to fail and the run to stop if screenscore was used. The error reported when this bug occurred was “Error! Screenscore::FindAcceptors (OH)”.

  • Fixed a crash bug when calculating MASC data for ligands on a 64bit machine.

  • Fixed minor when reporting how many molecules in a database have MASC data. The percentage reported was erroneously divided by 100.

  • Fixed a bug when using both the MASC and Non-MASC variants of a scoring function in the same run. The bug caused a shift in the Non-MASC score related to the precalculated MASC data, while the MASC variant score was correctly calculated. The error was especially damaging to CGT score.

  • Fixed spelling error by changing parameter -recalculate_masc_data to -recalculate_masc_data. The original misspelling is now an alias to minimize impact on users with existing parameter files.

  • Fixed bug checking for charges when -zapbind and -pro are used together. (Did not affect runs where -zapbind and -rec were being used). The bug caused the run to stop.

  • Fixed a bug in -clash_scale flag, that causes the value passed to the flag to be ignored and a value of one to be used. If the -clash_scale flag was not used no bug occurred in this regard.

  • Silenced the warning “OEInterface::Get, requesting value of unset parameter -addbox” when using the -box parameter without also specifying -addbox. The warning was spurious, no error occurred when it was issued.

  • Slave of a multiprocessor run now longer require a license, only the master process requires a license now.