2022 DOCK tutorial 2 with PDBID 4ZUD
- 1 Introduction
- 2 Protein and Ligand Preparation
- 3 Surface Spheres Generation
- 4 Grid Box
- 5 Footprint Analysis
- 6 DOCKING
- 7 Virtual Screening
- 8 Cartesian Minimization
- 9 Rescoring
DOCK is a molecular modeling program capable of sampling lower-energy ligand conformations with respect to a binding surface on a given protein. DOCK utilizes and manipulates the geometry of the ligand to find the conformation that yields that most favorable interaction with the respective binding site. With this tool, millions of molecules can be rapidly screened against a target protein for the purposes of identifying new drug molecules that are physiologically relevant.
For more information on DOCK and it's uses, please refer to their online manual: DOCK6 Manual
Before beginning, create the following directories in your space so that all necessary files are organized and can be access quickly:
mkdir 001.structure 002.surface_spheres 003.gridbox 004.dock 005.virtual_screen 006.cartesian_min 007.rescore
You don't have to name your directories the same as they are named here, but be cautious since the files that will be used for this tutorial utilize this naming scheme. They will need to be changed in each file that refers to them if you don't use this naming scheme!
Be sure to have Chimera installed on your system as it will be our primary visualization and system-editing program.
Protein and Ligand Preparation
Download the 4ZUD PDB file from the RCSB PDB website and open the file in Chimera.
Select -> Open -> (pathway to pdb file on your local machine) -> Open
You will notice a few side chain residues are explicitly displayed; those are the ones that directly engage with the ligand. The structure also has some missing regions denoted by the dashed-lines. These regions do not have to be modeled to use the system for docking since the majority of the protein remains restrained during the process (except for the residues of the active site, to a certain extent). You can play around with Chimera and visualize the protein from different angles to get a complete look at the protein to ensure there are no glaring errors in the structure that could have somehow arose from the downloading and opening process (Doesn't usually happen, but it's always good to be sure before moving on!)
Many structures deposited in the PDB lack hydrogens due to the difficulty in resolving their electron densities from cryo-EM or X-ray crystallography. The structures also lack formal charges since that information is not captured with out current experimental structure-determining techniques. Both charges and hydrogens are crucial for accurately studying any chemical system, and so they both must be added manually to 4ZUD in order to prime the system for docking.
First we want to delete the ligand from the file since we want to save the protein separately
Select -> Residue -> OLM Actions -> Atoms/Bonds -> Delete
4ZUD Structure Caveat
In the 4ZUD paper it mentions that ILE53 was mutated to an Alanine, but when you load the PDB into Chimera, it is recognized as an isoleucine but with an alanine side-chain.
There is no immediately obvious explanation for this, but since it is being recognized as an isoleucine, we're just going to edit the side chain to be the correct one. ILE53 can be selected by using the following command on the Chimera Command Line interface which could be accessed under Tools:
Tools -> General Controls -> Command Line
Make sure that you're selecting model 1 or whatever your current model is ranked at under Tools -> General Control -> Model Panel
Once residue 53 is selected, you can change the side-chain atoms by selecting a new rotamer type:
Tools -> Structure Editing -> Rotamers
In the Choose Rotamer Parameters window, select the ILE rotamer type and press apply.
In the ILE 53.A Side-Chain Rotamers window, choose the highest probability rotamers and press apply.
You should now see your alanine side chain change into an isoleucine one. It will remove the hydrogens so you have to add those back in, just for this residue.
To select the entire protein:
Select -> Chain -> A
To add the hydrogens to the protein:
Tools -> Structure Editing -> AddH
All residues in the protein should now have all of the hydrogens that were missing. Make sure to look at the output log of this command just in-case any errors arise, although there should be none if the instructions were followed thus far.
There is a similar Chimera command to add charges to your protein selection:
Tools -> Structure Editing -> Add Charge
After using this command you should receive an error stating that Correct charges are unknown for 3 non-standard atom names in otherwise standard residues. If you look at those atoms in the reply log, they're hydrogens belonging to ILE53. If you take a look at the paper that accompanies the 4ZUD structure, those hydrogens were replaced by tritium for crystallization purposes and Chimera does not recognize them as standard atoms and doesn't have predefined partial charges for them. Since Chimera doesn't recognize them, it will not apply charges to them. Additionally, since they are just hydrogens (meaning that their charge contributions to the system is often times very minimal) and the residue is not near the ligand active site, we do not have to do anything further in terms of adding charges.
The Add Charge command predicts the protein net charge to be -2.913. Make sure that this value is in agreement with what the PDB website or the corresponding paper says about the charge, if that information is provided.
Once you have completed all of the aforementioned steps, you have to save the protein as a mol2 file.
File -> Save Mol2
For the purposes of this tutorial the file will be called 4ZUD_protein_hydrogens.mol2
You will want to save a version of this file without hydrogens. To do that, you can select and delete all of the hydrogens like so:
Select -> Chemistry -> element -> H Actions -> Atoms/Bonds -> Delete
You can then resave the file. 4ZUD_protein_without_hydrogens.mol2
Now we want to focus on preparing the ligand. We can reopen the unmodified 4ZUD PDB in Chimera and delete the protein from the file. You can do this by doing an inverse selection for the protein. First select the ligand:
Select -> Residue -> OLM
On your keyboard, press Shift + Right-Arrow keys simultaneously to invert the selection to the parts that belong to the protein. You can then delete your selection, and you should be left with just the ligand:
Actions -> Atoms/Bonds -> Delete
To add hydrogens to the ligand, you can follow the same procedure that you used to add them to the protein § Adding Hydrogens.
You can follow the same procedures to add the charges to the ligand as well § Adding Charges.
The Add Charge command predicts a net -1 charge for the ligand. This makes sense since there is a carboxylate group in the molecules. Again, make sure to corroborate this with what the paper states and/or the PDB website states about the charge.
You can then save the file the same way you did for the protein mol2. For the purposes of this tutorial, the file will be named 4ZUD_ligand_hydrogens.mol2
Before going any further, make sure to copy the mol2 files that were generated thus far from your local computer to Seawulf. Place these files in the 001.structure directory, or whichever one is the equivalent directory in your records.
Surface Spheres Generation
UNDER CONSTRUCTION -- FILE NAMES, DIRECTORIES ARE FROM AS. WILL CHANGE THEM TO JA CONVENTIONS.
Next, we'll need to generate the grid we'll use in subsequent steps. Let's start by creating our grid.in file:
Into this file copy-paste the following:
compute_grids yes grid_spacing 0.4 output_molecule no contact_score no energy_score yes energy_cutoff_distance 9999 atom_model a attractive_exponent 6 repulsive_exponent 9 distance_dielectric yes dielectric_factor 4 bump_filter yes bump_overlap 0.75 receptor_file ../001_structure/4zud_receptor_wcharges_wH.mol2 box_file 4zud.box.pdb vdw_definition_file /gpfs/projects/AMS536/zzz.programs/dock6.9_release/parameters/vdw_AMBER_parm99.defn score_grid_prefix grid
Now that there is a grid.in file, the grid can be produced by running the following:
grid -i grid.in -o gridinfo.out
Upon successful completion of the grid generation, there should be three new files in your working directory: grid.bmp, grid.nrg, and gridinfo.out.
Before we do anything, we must be in the correct directory:
To generate a box around our subject, we need an .in file for Showbox:
Into which we will copy-paste the following:
Y 8.0 ../002_surface_spheres/4zud_receptor_trunc_spheres_sel.sph 1 4zud.box.pdb
Now that we have a .in file, we will produce a .box.pdb file by running the following:
showbox < showbox.in
After running this, you should have a 4ZUD.box.pdb file in the directory which you ran these commands.
While the ligand in the 4ZUD crystal structure seems to be in a reasonable pose (no steric clashes, etc.), and is of course bound to the receptor, it may not be representative of the lowest energy conformation. Assuming this is the case, the use of the non-minimized ligand conformation will reduce the accuracy of any calculations we perform with it.
One can avoid this problem by performing an energy minimization of the ligand. The first step to do this with 4ZUD will be to cd into the appropriate directory:
We'll be doing some other work in this directory later, so we should create a directory within this one specifically for performing our minimization:
cd into /min/ and produce an input file:
Into which the following should be copy-pasted:
conformer_search_type rigid use_internal_energy yes internal_energy_rep_exp 12 internal_energy_cutoff 100.0 ligand_atom_file ../001_structure/4zud_ligand_wcharges_wH.mol2 limit_max_ligands no skip_molecule no read_mol_solvation no calculate_rmsd yes use_rmsd_reference_mol yes rmsd_reference_filename ../001_structure/4zud_ligand_wcharges_wH.mol2 use_database_filter no orient_ligand no bump_filter no score_molecules yes contact_score_primary no contact_score_secondary no grid_score_primary yes grid_score_secondary no grid_score_rep_rad_scale 1 grid_score_vdw_scale 1 grid_score_es_scale 1 grid_score_grid_prefix ../003_gridbox/grid multigrid_score_secondary no dock3.5_score_secondary no continuous_score_secondary no footprint_similarity_score_secondary no pharmacophore_score_secondary no descriptor_score_secondary no gbsa_zou_score_secondary no gbsa_hawkins_score_secondary no SASA_score_secondary no amber_score_secondary no minimize_ligand yes simplex_max_iterations 1000 simplex_tors_premin_iterations 0 simplex_max_cycles 1 simplex_score_converge 0.1 simplex_cycle_converge 1.0 simplex_trans_step 1.0 simplex_rot_step 0.1 simplex_tors_step 10.0 simplex_random_seed 0 simplex_restraint_min yes simplex_coefficient_restraint 10.0 atom_model all vdw_defn_file /gpfs/projects/AMS536/zzz.programs/dock6.9_release/parameters/vdw_AMBER_parm99.defn flex_defn_file /gpfs/projects/AMS536/zzz.programs/dock6.9_release/parameters/flex.defn flex_drive_file /gpfs/projects/AMS536/zzz.programs/dock6.9_release/parameters/flex_drive.tbl ligand_outfile_prefix 4zud.lig.min
Now that we have our .in file, let's put it to use:
dock6 -i min.in -o min.out
Once this finishes running, two files will be generated: 4zud.lig.min_scored.mol2 and min.out. Copy 4zud.lig.min_scored.mol2 to your local machine and open it alongside the original ligand in chimera. It should something like this:
Where the blue structure is the ligand before minimization, and the beige ligand is after.
The main goal of this section is to conduct a pose reproduction of the crystal structure ligand to ensure that we are able to successfully re-produce the low-energy pose that the ligand was crystalized in. If docking is not successful at this, then this system may not be suitable for further testing using DOCK since it is not able to produce a pose that is already known to be sampled by the ligand, based on observations from the crystal. For these reasons, pose reproduction is a worthwhile preliminary step in any docking experiments so that we can be more confident in our results if the system will be used for more complex docking protocols such as virtual screening or de novo design.
In the current implementation of DOCK6, a sampling success is defined as a pose RMSD that is < 2 Å relative to the crystal pose.
Rigid docking is the least computationally intensive algorithm in the DOCK program since it does not allow for dihedral rotations, or bond length perturbations. Since we are just focusing on pose reproduction (and since we know that the crystal structure ligand pose is of biological relevance), this docking method would be a good first step.
In your 004.dock directory prepare this file:
Insert the following into that file:
conformer_search_type rigid use_internal_energy yes internal_energy_rep_exp 12 internal_energy_cutoff 100 ligand_atom_file ../003.gridbox/4zud.lig.min_scored.mol2 limit_max_ligands no skip_molecule no read_mol_solvation no calculate_rmsd yes use_rmsd_reference_mol yes rmsd_reference_filename ../003.gridbox/4zud.lig.min_scored.mol2 use_database_filter no orient_ligand yes automated_matching yes receptor_site_file ../002.surface_spheres/selected_spheres.sph max_orientations 2000 critical_points no chemical_matching no use_ligand_spheres no bump_filter no score_molecules yes contact_score_primary no contact_score_secondary no grid_score_primary yes grid_score_secondary no grid_score_rep_rad_scale 1 grid_score_vdw_scale 1 grid_score_es_scale 1 grid_score_grid_prefix ../003.gridbox/grid multigrid_score_secondary no dock3.5_score_secondary no continuous_score_secondary no footprint_similarity_score_secondary no pharmacophore_score_secondary no descriptor_score_secondary no gbsa_zou_score_secondary no gbsa_hawkins_score_secondary no SASA_score_secondary no amber_score_secondary no minimize_ligand yes simplex_max_iterations 1000 simplex_tors_premin_iterations 0 simplex_max_cycles 1 simplex_score_converge 0.1 simplex_cycle_converge 1.0 simplex_trans_step 1.0 simplex_rot_step 0.1 simplex_tors_step 10.0 simplex_random_seed 0 simplex_restraint_min no atom_model all vdw_defn_file /gpfs/projects/AMS536/zzz.programs/dock6.9_release/parameters/vdw_AMBER_parm99.defn flex_defn_file /gpfs/projects/AMS536/zzz.programs/dock6.9_release/parameters/flex.defn flex_drive_file /gpfs/projects/AMS536/zzz.programs/dock6.9_release/parameters/flex_drive.tbl ligand_outfile_prefix rigid.out write_orientations no num_scored_conformers 1 rank_ligands no
Run the input file with the following command:
dock6 -i rigid.in
Along with an out file, the rigid.out_scored.mol2 should have been created. This is your docked ligand(s). The ligand maybe be superimposed onto the crystal structure ligand using View Dock in Chimera:
Tools -> Surface/Binding Analysis -> ViewDock -> (Path to docked mol2) -> Dock 4, 5, or 6
The crystal reference ligand is colored in blue.
Notice how similar the reproduced pose is to the crystal pose. This is largely due to not allowing much position correction/movement during the docking; hence rigid docking. This is the simplest form of docking so you should expect to see very small RMSDs. The RMSD for this DOCK run is below our threshold so it is considered to be a sampling success.
Fixed Anchor Docking
With fixed anchor docking, we can introduce more flexibility into the conformational sampling when trying to re-dock the ligand into the active site. This is done by choosing certain fragments from the ligand as "anchors" which are manually placed in specific locations on the active site, which serve as points from where the other fragments that make up the known crystal ligand can attach to and grow from.
This docking algorithm will allow for the sampling of rotamers of the linker moieties attached to the fixed anchor(s), allowing for a more diverse free energy landscape to explore; possibly yielding more theoretically favorable ligand conformations!
In a new sub directory:
In a new input file:
Insert the following:
conformer_search_type flex write_fragment_libraries no user_specified_anchor no limit_max_anchors no min_anchor_size 5 pruning_use_clustering yes pruning_max_orients 10000 pruning_clustering_cutoff 100 pruning_conformer_score_cutoff 100.0 pruning_conformer_score_scaling_factor 1.0 use_clash_overlap no write_growth_tree no use_internal_energy yes internal_energy_rep_exp 12 internal_energy_cutoff 100.0 ligand_atom_file ../../001.structure/4ZUD_ligand_hydrogens.mol2 limit_max_ligands no skip_molecule no read_mol_solvation no calculate_rmsd yes use_rmsd_reference_mol yes rmsd_reference_filename ../../001.structure/4ZUD_ligand_hydrogens.mol2 use_database_filter no orient_ligand no bump_filter no score_molecules yes contact_score_primary no contact_score_secondary no grid_score_primary yes grid_score_secondary no grid_score_rep_rad_scale 1 grid_score_vdw_scale 1 grid_score_es_scale 1 grid_score_grid_prefix ../../003.gridbox/grid multigrid_score_secondary no dock3.5_score_secondary no continuous_score_secondary no footprint_similarity_score_secondary no pharmacophore_score_secondary no descriptor_score_secondary no gbsa_zou_score_secondary no gbsa_hawkins_score_secondary no SASA_score_secondary no amber_score_secondary no minimize_ligand yes minimize_anchor yes minimize_flexible_growth yes use_advanced_simplex_parameters no simplex_max_cycles 1 simplex_score_converge 0.1 simplex_cycle_converge 1 simplex_trans_step 1 simplex_rot_step 0.1 simplex_tors_step 10 simplex_anchor_max_iterations 500 simplex_grow_max_iterations 500 simplex_grow_tors_premin_iterations 0 simplex_random_seed 0 simplex_restraint_min no atom_model all vdw_defn_file /gpfs/projects/AMS536/zzz.programs/dock6.9_release/parameters/vdw_AMBER_parm99.defn flex_defn_file /gpfs/projects/AMS536/zzz.programs/dock6.9_release/parameters/flex.defn flex_drive_file /gpfs/projects/AMS536/zzz.programs/dock6.9_release/parameters/flex_drive.tbl ligand_outfile_prefix 4ZUD_fad write_orientations no num_scored_conformers 1 rank_ligands no
To run the analysis:
dock6 -i fixed.in -O fixed.out
The resulting docked pose:
The RMSD in this case is still small, but slightly larger than the rigid docking which is also evident in the image. It makes sense that the pose generated from fixed anchor docking is not as close to the crystal pose since we allowed the algorithm to explore different rotational conformations, increasing the likelihood of generating a new conformation.
In Flex Docking the ligand is fully flexible, as its name would suggest. All degrees of freedom are sampled and atoms have full rotational freedom about their bonds. Due to this rigorous approach, however, Flex Docking is computationally time-consuming. This of course means that we will need to submit our run to the queue. Before we do that, we need to make a .in file for our run: