DOCK 6.13 This is the release of DOCK6 which focuses on new methodologies for DOCK_DN allowing users to bias fragment selection [1]. Further, bug fixes and an additional descriptor scoring function were added. NEW IN DOCK 6.13 Biasing of fragments and torsions in DOCK_DN New methodologies have been added to DOCK_DN that allow for users to bias the selection of both fragments and torsions toward those of higher frequency in the provided set. This provides users with finer control over the frag and tors compositions of their final ensembles. Flex.defn nomenclature changes Changes to the flex.defn file have been introduced in this version. To ensure users can maintain compatibility with their current experiments, the DOCK6.11 flex.defn file has been renamed to flex_6_11.defn. MW Soft filter for MW filtering for DOCK_GA Previously, molecules generated by DOCK_GA could be filtered for their MW only using a hard cutoff. Now, similar to DOCK_DN, users are able to use a soft filter range for MW, allowing for some molecules above and below these cutoff values to be accepted. Ligand Efficiency added for Descriptor Score Users can now toggle the use of Ligand Efficiency when using Grid Score via Descriptor Score. This new feature can be used alongside Grid Score and will be calculated as: Ligand Efficiency = (Grid Score)/(# Heavy Atoms) Genetic Algorithm Users can now set desired relative frequencies of attempts for the different mutation types they have enabled for a given run. Miscellaneous Mutation Selection algorithm changes for DOCK_GA Mutation and fragment to mutate selection process has been changed so that the GA will not select a mutation type and then select a fragment incompatible with the selected mutation type. New Anchor & Grow Parameters The old parameter num_scored_conformers has been removed and replaced with two new parameters num_final_scored_poses and num_preclustered_conformers. Users can explicitly define how many final scored poses should be written out in A&G with num_final_scored_poses, and how many conformers are stored in memory with num_preclustered_conformers. This fixes an issue where less scored poses would be written than were requested if clustering was enabled. ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ BUG FIXES Descriptor score will now terminate if no valid scoring combination is selected. Since DOCK6.11, RDKit was implemented into DOCK_GA to compute cheminformatics descriptors. However, some offspring were missing these descriptors. This has been fixed; descriptors will now be included for all offspring. DOCK_GA's replacement mutation for scaffolds was performing fewer replacements than intended. This has been fixed. The flex.defn file has been updated to recognize N.pl3 and N.am atom types attached to S.o2 (attached to 2 O.2 and 1 O.3). The definitions added are: name sulfite-N_planar drive_id 3 minimize 1 definition S.o2 ( 2 O.2 [ * ] ) ( O.3 [ * ] ) definition N.pl3 ______________________________________________________________________ name sulfite-N_am drive_id 3 minimize 1 definition S.o2 ( 2 O.2 [ * ] ) ( O.3 [ * ] ) definition N.am ______________________________________________________________________ The mol2 file produced when setting write_conformers to yes during anchor & grow will no longer contain large quantities of duplicate conformers. Previously, anchor and grow would only hold a limited number of conformers in memory based on parameter num_scored_conformers. The New Anchor & Grow Parameters described above enable explicit control over this. Setting dn_write_pruned_duplicates to "no" will not suppress output from dn_write_prune_dump, and vice versa. Previously, verbose prune dump output in the de novo algorithm would only be written if both dn_write_pruned_duplicates and dn_write_prune_dump were set to yes. This behavior has been fixed, and the expected verbose files are now written for their respected option. ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ INPUT PARAMETER CHANGES AND THEIR DEFAULT VALUES Example: parameter_name [default_value] (options): Added: For Descriptor Score: descriptor_use_grid_lig_efficiency [no] (yes | no) descriptor_weight_grid_lig_eff [1] For DOCK_DN: dn_bias_with_fraglib [no] (yes | no) dn_use_roulette [no] (yes | no) ~with fraglib bias on~ dn_sel_frag_by_freq [no] (yes | no) dn_accept_frag_by_freq [no] (yes | no) For DOCK_GA: ga_mol_wt_cutoff_type [hard] (hard | soft) ga_constraint_upper_mol_wt [500.0] ga_constraint_lower_mol_wt [0] ga_constraint_mol_wt_std_dev [35.0] For any run involving Anchor & Grow: num_final_scored_poses [1] num_preclustered_conformers [num_final_scored_poses] ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ OUTPUT COMPONENTS ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ KNOWN BUGS 1. When selecting both internal energy and internal energy score, the parameter "internal_energy_rep_exp" appears twice in the question tree, and it will be queried during user input when DOCK is invoked from the command-line even if the value is already set. 2. Scoring functions are only able to handle receptors with less than 2000 atoms. It's possible to use more, but that may introduce undefined behavior in the current implementation. If users wish to use more, they should contact the developers. 3. In the flex.defn file, there is a flag minimize which is disabled. (The flag has no impact on the calculation.) All torsions are minimized. 4. RDKit and thus dock6.rdkit executables can be difficult to install on non Linux platforms, such as macOS and Cygwin. Users should consult the DOCK-Fans mailing list. ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ RELEASE DATE July 2025 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ REFERENCES (1) Bickel, J. D., Boysan, B. T., and Rizzo, R. C. (2025) Fragment and torsion biasing algorithms for construction of small organic molecules in proteins using DOCK. J. Comput. Chem. 2025 46, e27508. DOI: 10.1002/jcc.27508