OEDocking 3.0.0¶
April 2012
Changes¶
The OEDocking 3.0.0 release is the first release of the OEDocking package, which is an upgrade of the previous FRED release packages. The OEDocking package includes FRED along with a suite of new docking tools.
As part of this OEDocking release the interface to FRED has been streamlined. Several functions of FRED have been split out into separate programs that are now part of the OEDocking suite, and several older features that have not proved to improve docking results (or in the worst case degrade docking results relative to the defaults) have been removed.
The features of FRED 2.2.5 that have been moved to separate utility programs in 3.0.0 are as follows
Receptor Creation
Receptors are now created either with the MakeReceptor GUI or the
pdb2receptor
,apopdb2receptor
, orreceptor_setup
command-line programs. Existing receptors can now also be edited or modified with the MakeReceptor GUI or ReceptorToolbox command line program. These programs are all included in the OEDocking distribution.Relevant flags removed from FRED:
-pro
-strip_water
-bound_ligand
-box
-addbox
-no_inner_contour
Hybrid Docking
Hybrid docking is now available as a separate application, HYBRID, included in the OEDocking distribution.
Rescoring of poses
The feature to rescore existing poses is now available in a separate application, ScorePose, included in the OEDocking distribution.
The features of FRED 2.2.5 removed in 3.0.0 are as follows
Alternate Scoring Functions
FRED now uses the improved Chemgauss4 scoring function. This scoring function has better virtual screening and pose prediction performance than any of the scoring functions available in FRED 2.2.5.
Relevant flags removed from FRED:
-exhaustive_scoring
-opt
-shapegauss
-plp
-chemgauss2
-chemgauss3
-chemscore
-oechemscore
-screenscore
-cgo
-cgt
-zapbind
-consensus
-shapegauss_masc
-plp_masc
-chemgauss2_masc
-chemgauss3_masc
-chemscore_masc
-oechemscore_masc
-screenscore_masc
-cgo_masc
-cgt_masc
-zapbind_masc
-consensus_masc
-assign_ligand_charges
MASC
Multiple Active Site Correction (MASC) was introduced in FRED 2.1 as a method of compensating for the size bias in scoring functions. This size bias arises in many scoring functions where the interaction terms are all favorable interactions, and thus larger molecules score better since they can make more interactions. Chemgauss4 (and Chemgauss3), however, has both favorable interactions (i.e., shape, hydrogen bonding and metal-chelator) and unfavorable interactions (i.e., desolvation and clash), and therefore does not have a significant size bias that necessitates the use of MASC.
Relevant flags removed from FRED:
-reference_receptors
-no_masc_data
-recalculate_masc_data
-report_masc_failures
Consensus Pose Selection
FRED 2.2.5 had slightly better pose prediction results when the best docked pose was selected using a consensus of three scoring function (Chemgauss3, Chemscore and PLP) rather than one (Chemgauss3). The new Chemgauss4 scoring function in the 3.0 release is not improved by this feature, and has therefore been removed.
Relevant flags removed from FRED:
-pose_select_weight_shapegauss
-pose_select_weight_plp
-pose_select_weight_chemgauss2
-pose_select_weight_chemgauss3
-pose_select_weight_chemscore
-pose_select_weight_oechemscore
-pose_select_weight_screenscore
-pose_select_weight_cgo
-pose_select_weight_cgt
MMFF Refinement
Using this option in FRED 2.2.5 degraded the results for both pose prediction and virtual screening and significant increases the run time.
Relevant flags removed from FRED:
-refine
New Features¶
Chemgauss4 scoring
Improved hydrogen bond detection vs. Chemgauss3. Hydrogen bond network effects are also now accounted for.
Hybrid docking
Dock using ligand and protein structural information simultaneously
MPI
MPI is now supported on all platforms except Windows.
Docking Report
Creates an Adobe PDF docked report for selected docked molecules.
Receptor tools
Streamlines the receptor setup GUI, and also added several command line tools for setting up receptors.
Bug Fixes¶
Fixed a bug where large hitlists (> ~50,000) would result is very slow runtimes.
SD tag data on the input ligand is now retained.
Added an option to toggle hardware rendering in the make_receptor GUI application to correct Windows driver related 3D graphics issues.