MovableType Tutorials

MovableType Tutorials

Have questions? Explore MovableType FAQs or Contact Us!

Note: These tutorials have been updated for DEV.671 or later. If you do not have a current version, please email sales@quantumbioinc.com for download instructions. The MovableType or MT Method is a fast, free energy method based on recent work by Drs. Zheng Zheng and Kenneth M. Merz Jr. at Michigan State University. This patented method was licensed from MSU by QuantumBio specifically to address the needs of the pharmaceutical industry (please see the literature for more information). The product is broken down into several key modules for small molecule or ligand Conformational Search (MTCS), ligand docking (MTDock), EndState protein:ligand affinity prediction (MTScoreES) and Ensemble protein:ligand affinity prediction (MTScoreE), and finally protein loop and rotomer modeling (MTFlex). Before moving forward, if you have not already done so, you should review the Installation Instructions for the DivCon Discovery Suite (which includes MovableType). The current MT modules include MTScoreES, MTScoreE, MTDock, and MTCS. MTFlex will be added in the coming months as it comes on line, so please keep an eye on this website and Contact Us if you would like to consider the use of the technology in your own site. The tutorials currently available include:

Structure Preparation & Scaling

Structure preparation for MT-based methods is similar to preparation for QM-based methods or molecular dynamics simulations. There are various 3rd party graphical user interface (GUI) platforms on the market. QuantumBio software has been used most with the MOE platform from Chemical Computing Group; however, you may try other platforms as well and generally these steps are transferable to analogous tools in Maestro, Discovery Studio, Sybyl, and so on. If you have any trouble, Contact Us for help and recommendations. For the MT tool you will need to perform a number of standard preparation steps in a GUI. These preparation steps are noted below, and the details of the MOE structure preparation toolbox are available under: MOE -> Help -> Contents -> 4. Data Preparation -> “Prepare a Protein or Protein-Ligand Complex”

  1. (Required) Structure Assessment: The overall quality of your structural model will likely be the best driver of success of the MT process. The principle of “Garbage In / Garbage Out” is very much a fundemental premise in all simulations (including MT as well as MD). Generally, the higher (smaller) the resolution of any X-ray structures, the agreement of the model with the density, and the completeness of the structure (especially in the active site) are all important. Likewise, you may need to review alternative structural data (alternative atom positions). Finally, explicit water molecules should be reviewed: MT will use any explicit water molecules (such as bridging waters) in the calculation and these waters will be included as protein “residues” in the protein and protein:ligand partition functions. Therefore, any water molecules you do not wish to include should be removed.
  2. (Required) Protonation: Unfortunately, many X-ray crystal models are devoid of protons (even key protons which are used in protein:ligand binding). MT is what is called an “all atom” method so you must add protons and you must add them carefully. MOE includes the Protonate3D tool which will add protons and minimize the H-bonding network, and allow you to visualize those protons to help double check their addition. The Protonate3D tool, along with numerous other structure preparation tools and indicators, are available under the Structure Preparation Compute Menu: MOE -> Compute -> Prepare -> “Structure Preparation…”
  3. (Recommended) Structure Minimization or refinement: Any added atoms (protons, waters, rotomers, and so on) should be minimized using MOE’s built-in structure minimization tool. And depending upon the quality of the starting X-ray model, some structure minimization or refinement will need to be performed on the entirety of the structure. In order to maintain the active site, generally any minimization should be performed with a bound ligand left in place (otherwise minimization can “collapse” the active site).
  4. (Recommended) Ligand Docking: The MT method consists of two scoring types including MTScoreES [EndState] scoring and MTScoreE [Ensemble] scoring. Both scoring types include some sampling. The difference is the EndState score only includes “smeared” or “blurred” sampling around a single protein:ligand pose. Ensemble scoring on the other hand uses the same blurred sampling, but applies it to a set of provided poses. These poses can be thought of as landscape minima as discussed by Mobley and Dill and generally all of them are included in the MTScoreE [Ensemble] score. Therefore, you should attempt to include only docking poses which you feel accurately represent the potential binding modes of the structure. As a general “rule of thumb,” as reviewed on the FAQ, 10-15 poses are adequate with DEV.671 to reach convergence. You may use either induced-fit or rigid-target docking and you should review the linked tutorials to understand the differences. MOE includes a Docking Panel which can be used to set up and run the docking process or you may use the qbDockPair.svl script enclosed in the DivCon Discovery Suite package. To use the GUI (and to address cases which may be outside of the limits of a script), you should review the MOE documentation and tutorials for this docking by visiting MOE -> Help -> Contents -> “8. Docking” and review the courses and tutorials available on the CCG Support website.
  5. (Required) Execute MTScoreE [Ensemble]: Finally, once the docking process is complete, will need to choose the tutorial below which is appropriate to the type of docking you chose to perform above. These generally fall into two, supported types of docking:
    1. Rigid-Target Docking: As a first attempt, docking is performed using a rigid-protein / flexible ligand regime. The 2-step Rigid-Target Tutorial on this page is provided to demonstrate this workflow.
    2. Induced Fit Docking: Support for induced-fit docking was just added to DivCon Discovery Suite DEV.671. Within induced fit docking, the target is optimized for each ligand pose, and this configuration is often a more accurate representation. Note: if you run this workflow, then you should be careful to also optimize the bound protein:ligand wildtype complex you prepared during step 3 above. The Induced Fit Tutorial on this page is provided to demonstrate this workflow

If you require support for the use of MOE, you may Contact Us or visit the CCG Support page. A Note On Scaling: As with any modeling or simulation method, some scaling of final results is expected in order to produce ∆G values with low absolute errors and some scaling should be performed in your own work based on a set of knowns. Generally, scaling does not impact correlation but it does impact total error (such as mean unsigned error or MUE or RMSE). By default, all values produced by MTScoreE and MTScoreES are scaled according to our treatment of 795 chosen PDBBind structures and most of our reported results are based upon this default scaling. However, as one gains a better understanding of the structures within a project, using a leave one out (LOO) or other analogous process is suggested. In a project-driven pharmaceutical research endeavor, you may want to choose a set of scaling factors which are determined retrospectively but are then applied prospectively.

Shell Environment

The shell (bash, tcsh, etc) environment is currently used to communicate certain settings to the MovableType (MT) modules within the DivCon Discovery Suite. These options will be folded into the software in the not too distant future as executable options, but currently the following environment variables can be set prior to running the MT calculations:

  • The MT_NOSCALE environment variable can be used to disable the “built in” linear-scaling in the MTScoreES and MTScoreE software. The default scaling is based on the characterization of 795 reference compounds from the PDBbind set. Note: both the scaled [MTScore*(s)] and unscaled versions of all calculations are provided. Scaling does not impact correlation with experiment.
    bash: % export MT_NOSCALE=1
    
    csh/tcsh: % setenv MT_NOSCALE 1
  • The MT_SKIP_H2O environment variable can be used to disable the inclusion of water molecules in the MTScoreES and MTScoreE simulation. The default (e.g. unset MT_SKIP_H2O) will include water and any other atoms, salts, cofactors, and so on included in the system.
    bash: % export MT_SKIP_H2O=1
    
    csh/tcsh: % setenv MT_SKIP_H2O 1

Command Line: MTConfSearch

The MTCS algorithm uses the same data developed for the MTScore algorithm in order to determine the most likely conformers of a ligand. This algorithm is often used as part of the MTScore tool (for sampling) but it can also be used independently. The method can take either a mol2 file or a SMILES string and generate the associated set of conformers (as an SDF). In this example, the SMILES string is provided and all 30 unique conformers are written to the SDF using the -p (for publish) command line argument.

/path/to/DivConDiscoverySuite/bin/qmechanic c1ccccc1CCCC --mtcs -p sdf

Command line:
qmechanic c1ccccc1CCCC --mtcs
Version: DivCon Discovery Suite


Running on HOSTNAME
Processors: 1
# CONFORMER AFTER FIRST RUN: 36
TIME AFTER CONFORMER GENERATION: 0.003246 depth: 3

In addition to the SMILES approach, you can also provide a mol2 file with the associated ligand. Whether the input is SMILES or mol2, it should work the same and provide similar output.

/path/to/DivConDiscoverySuite/bin/qmechanic butylbenzene.mol2 --mtcs -p sdf

Command line:
qmechanic butylbenzene.mol2 --mtcs 
Version: DivCon Discovery Suite 


Running on HOSTNAME
Processors: 1
# CONFORMER AFTER FIRST RUN: 36
TIME AFTER CONFORMER GENERATION: 0.003246 depth: 3

Additional Examples

Command Line: MTScore (Endstate)

The MTScore (EndState) or MTScoreES method is a fast free energy method which determines the end state binding affinity of a single conformer of a protein:ligand complex. Unlike with the ensemble score (or MTScoreE) the MTScoreES method only scores the final configuration and pose as provided, and gives one a quick understanding of whether that pose is representative. As with MTConfSearch, MTScoreES is a fast and easy calculation to perform and it is highly scriptable in the command line. Currently, you must separate the protein from the ligand and provide these two items in two, separate files, but this will change in the not too distant future to allow you to “select” the ligand within a PDB. Correct protonation is suggested in order to aid in the automatic atom typing process.

/path/to/DivConDiscoverySuite/bin/qmechanic 4w7t_protein.pdb --ligand 4w7t_ligand.mol2 --mtscore endstate

Command line:
qmechanic 4w7t_protein.pdb 4w7t_ligand.mol2 --mtscore endstate
Version: DivCon Discovery Suite 1 DivConDiscoverySuite-7.5.0-b4411


Running on HOSTNAME
Processors: 1
----------------------------------- MTScore ------------------------------------
"Species" "Interact" "E_Sol(complex)" "E_Sol(ligand)" "E_Sol(protein)"
--------------------------------------------------------------------------------
"4w7t_ligand" -232.787273 -6.310488 -8.545937 -13.371086
--------------------------------------------------------------------------------


----------------------------------- MTScore ------------------------------------
"Species" "MTScoreES"
--------------------------------------------------------------------------------
"4w7t_ligand" -9.066661
--------------------------------------------------------------------------------



Job Complete
Total Computation Time (Seconds): 6.29878
Fri Apr 27 16:33:59 2018

In this output, the MTScoreES (endstate) is written out in the final table. The previous table has various additional terms (including the gas phase interaction energy or “Interact.”

Additional Examples

MOE GUI – MTScoreES (Endstate)

The MTScore (EndState) or MTScoreES method is a fast free energy method which determines the end state binding free energy of a single conformer of a protein:ligand complex. Unlike with the ensemble score (or MTScoreE) the MTScoreES method only scores the final configuration and pose as provided and gives one a quick understanding of whether that pose is representative. The MOE interface to MTScoreES is available through the use of the new qbWebServcice module. The following command is called prior to running MOE, and the 8080 port is provided as an argument in order for the webservice to “listen” on the 8080 port. Once you have started the webservice, you should take note of the hostname or IP address of the computer running the service. If you started the webservice on the same computer which will be running MOE, then this hostname would generally be “localhost” – however, if you have started the webservice on a separate, remote computer, you will need the name of the computer on your network. Ask your system administrator for more information if required.

/path/to/DivConSuite/bin/qbwebservice 8080

The MTScoreES tool tray in MOE allows you to work with MTScoreES in an approximately “real time” manner. The try can be paired with the built-in MOE molecule builder, structure preparation tools, molecular database (MDB) project format, and of course the molecular visualizer.

  1. In order to start the QuantumBio/MTScore tool tray, after you have started a webservice, simply go to the Extra->QuantumBio->MTScoreES Interactive menu. This will instantiate the tray on the right side of the screen. At first, if there is no structure loaded into MOE, the “Status” for the widget will show you that you will need to open a structure of your choosing. This Status text will appear and update as you prepare the strucuture for treatment. Use standard CCG/MOE tools (e.g. Structure Prep, Protonate3D, etcetera) to prepare the structutre for treatment. Likewise, you should click the “gears” button next to the qbWebService text to set your hostname and port as per the above noted step.
  2. Once the structure is ready, you will be required to open or create an MDB (molecule database) through the interface using the “open folder” button. This file will act as the primary storage mechanism for molecular data, scores, labels and so on. This file allows you to store your work from one session to the next, and it will also allow you to pass your data to others in your firm or group. You may use a previously created MDB file, or you may create a new one.
  3. The “Add” button is used to score the currently viewed structure using the qbWebService and add it to the list (and to the MDB). Prior to scoring, you may choose to optimize the Ligand in the pocket (using the MOE-chosen potential), the Pocket as a whole inclusive of the ligand, or None. Finally, the “X” button will remove the chosen structure from the list. If you accidentally delete a structure, you can always hit the Add button again and the webservice will re-score it. By default the current timestamp is used as a Label, but you may Rename or change the Label as you see fit. This Label is for your own use and is not used anywhere else in the calculation.If you open a previously generated MDB file, as in the screenshot below, you will see the list of poses/complexes. You may merge, delete, edit the MDB file using standard MOE tools, and MOE will update the list on the Tray automatically.

Additional Examples

 

MOE GUI – MOE Dock (Endstate)

After the qbWebService is running, you may start MOE and read in a protein and a ligand as per the CCG documentation. You may wish to familiarize yourself with the MOE documentation for the use of the Dock Panel. This documentation is available in your MOE installation at the following URL:

file:///$MOE/html/apps/docking.htm#Using_the_Dock_Panel

When you run the MOE and the above noted CCG-provided tutorial, if you have properly installed the QuantumBio plugins to MOE, you should see an option such as the one circled in red below. You may choose this option during the tutorial to observe the execution in action. The resulting MTScore will be provided as the “S” Score in the dock.mdb file and the MTScoreES [QuantumBio] option can be chosen anywhere scores are used in MOE.

Additional Examples

 

“3Step” MTCS+MTScoreE (Ensemble) with an External Docker

The MTScoreE (Ensemble) version of MTScore requires an ensemble of ligand poses to be generated and scored in order to sample the active site and accurately calculate a binding affinity. Generally, the better the poses the more accurate the affinity prediction, but the method can be quite forgiving. It is therefore suggested that you provide 25 to 50 poses per ligand. Since each docking package (e.g. MOE, GLIDE, etcetera) is different and each has its own strengths and weaknesses, MTScore can be used to score an ensemble of poses provided by an external docking program as well as its own built-in Heatmap docking approach (MTDock). In this example, we demonstrate the use the docking algorithm found within MOE. The concept though should be the same regardless of package (as long as the package can use and preserve the conformers provided by MTCS (Conformational Search). The following steps are performed: The input/output associated with this calculation is available in the Bace_030215_CAT_4p.tar.gz file.

  1. Set several environment variables which are used to manipulate the calculation.
    export DIVCON_INSTALL=/path/to/DiscoverSuite
  2. Create an empty directory and copy the input files (pdb and mol2) into the directory.
    mkdir BACE ; cd BACE ; cp ../Bace_030215_CAT_4p.tar.gz . ; tar zxvf Bace_030215_CAT_4p.tar.gz
    cp Bace_030215_CAT_4p/Bace_030215.pdb .
    cp Bace_030215_CAT_4p/Bace_030215_CAT_4p.mol2 .
  3. Use MTConfSearch (–mtcs) to generate a set of conformers based upon the ligand provided. Note that both the target and the ligand are taken as input. This convention is in order to correctly generate the target.h5 file which will be automaticailly read by qmechanic in step 5. In addition to the target.h5 file, the ligandname_conf.sdf file – including the generated MTCS conformers – is also produced using the –mtcs command line argument followed by the -p arguement (for publish). This ligandname_conf.sdf file will be used in step 4.
    ${DIVCON_INSTALL}/bin/qmechanic Bace_030215.pdb --ligand Bace_030215_CAT_4p.mol2 --mtcs -p sdf --np 4 -v 2
  4. Use MOEbatch to run a docking calculation with the resulting conformer SDF file using the SVL script provided with the DivConSuite. In this case, the receptor pdb file is provided using the -rec argument, the -lig argument is the ligand to be docked, and the -conf argument is the above noted ligand_conf.sdf file. This step will create a file called ligandname_dock.sdf which will be used in step 5.
    moebatch -licwait -run "${DIVCON_INSTALL}/svl/run/qbDockPair.svl" -rec Bace_030215.pdb -lig Bace_030215_CAT_4p.mol2 -conf Bace_030215_CAT_4p_conf.sdf -delwat
    Note: if you would wish to substitute your own docking software (e.g. GLIDE, GOLD, FlexX, etcetera) for this step you may do so. Be sure to configure the software to maintain/keep the conformers and SDF tags generated by MTCS in step number 3 above so that these tags may be used in step 5 below. MOE maintains these tags by default, and the above noted MOE/SVL script uses the MTCS conformers as the docked conformers. The script also will optimize each provided pose using the chosen potential. If you do substitute your own docking software (GLIDE, etcetera) in place of MOE, you should be familiar with the software and the same “garbage-in/garbage-out” principle applies. If you would like us to review your GLIDE (or MOE) generated input, please Contact Us and we’ll be happy to schedule a WebEx to review it.
  5. Finally, run MTScore on the ensemble in order to calculation the ensemble score. In this example, the –mtscore input is the MOE-generated dock file while the other two input files are the same target.pdb and ligand.mol2 files used above.
    ${DIVCON_INSTALL}/bin/qmechanic Bace_030215.pdb --ligand Bace_030215_CAT_4p.mol2 -h garf --mtdock Bace_030215_CAT_4p_dock.sdf --mtscore ensemble --np 4 -v 2
    ---------------------------------- Pose Score ----------------------------------
     S Pose  RMSD      ZPt(i)    ZPnb(i)     ZLt(i)    ZLnb(i)   ZPLnb(i)     dGsolv  MTScoreES MTScoreES(s)
    --------------------------------------------------------------------------------
       0      1.01    -698.37   -8994.09    -248.66   -1124.68  -15689.95       7.90  -15682.04      -9.51
       9      7.44    -698.37   -8994.09    -246.35   -1123.41  -15542.70       7.89  -15534.81      -9.46
       1      1.52    -698.37   -8994.09    -249.35   -1124.55  -15538.18       8.70  -15529.49      -9.46
       5      1.49    -698.37   -8994.09    -246.31   -1124.64  -15434.97       8.23  -15426.74      -9.42
       24     2.18    -698.37   -8994.09    -238.89   -1124.07  -15355.09       8.81  -15346.28      -9.40
       16     8.04    -698.37   -8994.09    -246.42   -1122.59  -15235.04      10.09  -15224.95      -9.35
       19     7.32    -698.37   -8994.09    -250.15   -1124.36  -15135.24       9.51  -15125.73      -9.32
       6      7.10    -698.37   -8994.09    -245.27   -1122.72  -15071.42       9.24  -15062.17      -9.30
       20     8.53    -698.37   -8994.09    -245.42   -1124.40  -15066.58       9.86  -15056.72      -9.30
       2      7.56    -698.37   -8994.09    -244.15   -1124.91  -15048.72       9.13  -15039.59      -9.29
       22     7.69    -698.37   -8994.09    -251.27   -1124.65  -14979.03      10.27  -14968.76      -9.27
       8      7.65    -698.37   -8994.09    -249.96   -1124.45  -14923.73       9.20  -14914.53      -9.25
       21     7.70    -698.37   -8994.09    -252.75   -1124.18  -14862.78      10.06  -14852.72      -9.23
       15     7.69    -698.37   -8994.09    -250.42   -1124.32  -14816.44      10.27  -14806.16      -9.21
       13     7.24    -698.37   -8994.09    -244.73   -1121.97  -14804.28       8.50  -14795.78      -9.21
       4      7.77    -698.37   -8994.09    -250.41   -1124.24  -14669.82       9.26  -14660.56      -9.16
       17     8.00    -698.37   -8994.09    -245.60   -1123.82  -14629.47       9.25  -14620.21      -9.14
       3      8.59    -698.37   -8994.09    -246.52   -1124.40  -14544.13       9.02  -14535.12      -9.12
       12     8.43    -698.37   -8994.09    -245.86   -1124.21  -14517.25       9.28  -14507.97      -9.11
       11     7.90    -698.37   -8994.09    -250.78   -1124.28  -14455.24       9.81  -14445.43      -9.08
       23     7.63    -698.37   -8994.09    -244.94   -1122.06  -14277.58       8.88  -14268.70      -9.02
       7      7.82    -698.37   -8994.09    -251.25   -1124.59  -14238.34       9.56  -14228.78      -9.01
       10     4.14    -698.37   -8994.09    -243.01   -1123.87  -14098.83       8.57  -14090.26      -8.96
       18     6.79    -698.37   -8994.09    -250.02   -1124.55  -13547.20      10.85  -13536.35      -8.77
       14     6.69    -698.37   -8994.09    -252.52   -1124.56  -13523.46      10.77  -13512.68      -8.76
    --------------------------------------------------------------------------------
    Status:
            * Pose not included in MTScoreE Calculation due to close rmsd to another pose.
    
    
    -------------------------------- Final MTScore ---------------------------------
    Species              wtMTScoreES  Z_PLnbsol       Z_PL        Z_L        Z_P   MTScoreE MTScoreE(s)
    --------------------------------------------------------------------------------
    Bace_030215_CAT_4p        -9.53  -15682.04  -26747.85   -1382.21   -9692.47  -15673.17      -9.51
    --------------------------------------------------------------------------------
    

     

Additional Examples

 

“2Step” MTScoreE with an External Docker (no-MTCS)

As noted in the Tutorial above, generally in order to correctly calculate the binding affinity, ∆G, of a protein:ligand complex one is expected to calculate both the complex partition function, ZPL, and the ligand partition function, ZL. This goal is usually accomplished by using both –mtcs and –mtscore. However, success of the above workflow rests on the ability of the docking software to maintain the same conformers provided by MTCS, and the score tags written out by MTCS during the conformer generation step. Some docking functions do not do that, and likewise, they often significantly impact the provided conformers through optimization, flip-state exploration, and other means. With the release of DEV.663, MTScoreE no longer requires the ZL term and the software can calculate the intra-molecular partition function internally. The benefit of this “2Step” approach is that your docking software (e.g. GLIDE, MOE, etcetera) is no longer biased using MTCS and therefore if a docking package is better suited for a particular protein structure, MTScoreE will still work as well as or better than the “3Step” MTScoreE workflow above. The MTScoreE (Ensemble) version of MTScore requires an ensemble of ligand poses to be generated and scored in order to sample the active site and accurately calculate a binding affinity. Generally, the better the poses the more accurate the affinity prediction, but the method can be quite forgiving. It is therefore suggested that you provide 25 to 50 poses per ligand. The following steps will allow you to run the “2Step” workflow:

  1. Set several environment variables which are used to manipulate the calculation.
    export DIVCON_INSTALL=/path/to/DiscoverSuite 
  2. Create an empty directory and copy the input files (pdb and mol2) into the directory.
    mkdir BACE ; cd BACE ; cp ../Bace_030215_CAT_4p.tar.gz . ; tar zxvf Bace_030215_CAT_4p.tar.gz
    cp Bace_030215_CAT_4p/Bace_030215.pdb . 
    cp Bace_030215_CAT_4p/Bace_030215_CAT_4p.mol2 . 
  3. Use MOEbatch to run a docking calculation with the resulting conformer SDF file using the SVL script provided with the DivConSuite. In this case, the receptor pdb file is provided using the -rec argument, the -lig argument is the ligand to be docked, and the -conf argument is the above noted ligand_conf.sdf file.
    moebatch -licwait -run "${DIVCON_INSTALL}/svl/run/qbDockPair.svl" -rec Bace_030215.pdb -lig Bace_030215_CAT_4p.mol2 -delwat
    Note: you may substitute your own docking software (GLIDE, etcetera) in place of MOE, but you should be familiar with the software and the same “garbage-in/garbage-out” principle applies. If you would like us to review your GLIDE (or MOE) generated input, please Contact Us and we’ll be happy to schedule a WebEx to review it.
  4. Finally, run MTScore on the ensemble in order to calculation the ensemble score. In this example, the –mtscore input is the MOE-generated dock file while the other two input files are the same target.pdb and ligand.mol2 files used above.
    ${DIVCON_INSTALL}/bin/qmechanic pro_Bace_030215_CAT_4p_predock.pdb --ligand lig_Bace_030215_CAT_4p_predock.mol2 -h garf --mtdock Bace_030215_CAT_4p_dock.sdf --mtscore ensemble --np 4 -v 2

Additional Examples

MTScoreE with induced fit protein:ligand docking

We know that binding often involves not only the movements of the ligand within the active site, but the movements of the active site around the ligand. Whether we’re talking about the “minimization-driven” shifts of protein residues to accomodate ligand conformers, rotomer flips or even larger loop movements, a true determination of binding affinity requires that we better represent these target as well as ligand transformations. With DEV.671, you may provide multiple PDB files (one for each protein:ligand conformer) in place of the –mtdock *_dock.sdf file noted above. These *_dock.pdb files should include one protein:ligand “snapshot” per file along with any critical water molecules and other cofactors. The software will warn you if you include an unsupported atom type, and – when using MT-AMBER – electrostatics will be calculated using the PM6 Hamiltonian. Note: use of induced fit or dynamic protein requires the use of the “2Step” method above. The following steps will allow you to run the “2Step” workflow:

  1. Set several environment variables which are used to manipulate the calculation.
    export DIVCON_INSTALL=/path/to/DiscoverSuite 
  2. Create an empty directory and copy the input files (pdb and mol2) into the directory.
    mkdir BACE ; cd BACE ; cp ../Bace_030215_CAT_4p.tar.gz . ; tar zxvf Bace_030215_CAT_4p.tar.gz
    cp Bace_030215_CAT_4p/Bace_030215.pdb . 
    cp Bace_030215_CAT_4p/Bace_030215_CAT_4p.mol2 . 
  3. Use MOEbatch to run a docking calculation with the resulting conformer SDF file using the SVL script provided with the DivConSuite. In this case, the receptor pdb file is provided using the -rec argument, the -lig argument is the ligand to be docked, and the -conf argument is the above noted ligand_conf.sdf file.
    moebatch -licwait -run "${DIVCON_INSTALL}/svl/run/qbDockPair.svl" -rec Bace_030215.pdb -lig Bace_030215_CAT_4p.mol2 -inducedfit -delwat
    Note: you may substitute your own docking software (GLIDE, etcetera) in place of MOE, but you should be familiar with the software and the same “garbage-in/garbage-out” principle applies. If you would like us to review your GLIDE (or MOE) generated input, please Contact Us and we’ll be happy to schedule a WebEx to review it.
  4. Finally, run MTScore on the ensemble in order to calculation the ensemble score. In this example, the –mtscore input is the MOE-generated dock files while the other two input files are the same target.pdb and ligand.mol2 files used above.
    ${DIVCON_INSTALL}/bin/qmechanic pro_Bace_030215_CAT_4p_predock.pdb --ligand lig_Bace_030215_CAT_4p_predock.mol2 -h amberff14 --mtdock *_dock.pdb --mtscore ensemble --np 4 -v 2

Additional Examples

Manipulating cutoffs to maximize predictions

The settings for the MTScoreE and MTScoreES methods are fairly limited in part because success with the method is usually ultimately dependent upon the quality and size of the protein:ligand ensemble. The one area where some manipulation is possible which can lead to better results is in how the pocket is defined and how non-bonded cutoffs are applied. With the MT method, you don’t need the pocket to cover the entire structure and in fact often the inclusion of too much protein will “dampen” the impact of the ligand in the ligand binding calculation. This is likely an artifact of the method and one which is being actively explored. With that in mind, we have two ways to manipulate the cutoffs: the non-bonded atomic pair function cutoff and the pocket definition:

  • –nb-cutoff XX: Adding the –nb-cutoff XX command line argument and value to any scoring process will “throw out” any atom-atom non-bonded pairs beyond XX Angstroms. The DEFAULT value at this time is 11.0 Å.
  • –ligand placed_ligand.mol2 XX or –pocket placed_ligand.mol2 XX : The –ligand and the –pocket command line arguments are synonyms. The –ligand argument has always been used to define the pocket with regards to the position of the species found in the placed_ligand.mol2 file. The XX number in Angstroms can now be added in order to this command line in order to define the size of the pocket (residue extended) beyond the placed_ligand. The DEFAULT value at this time is 8.0 Å.

As an example, imagine the MTScoreE [Ensemble] calculation calls in the tutorials above. The following settings will change the above noted DEFAULT values to 8.0 for the non-bonded cutoff and 4.5 for the pocket cutoff. These values will often yeild good results for certain sets:

 ${DIVCON_INSTALL}/bin/qmechanic Bace_030215.pdb --ligand Bace_030215_CAT_4p.mol2 4.5 -h amberff14 --mtdock *_dock.pdb  --mtscore ensemble --nb-cutoff 8.0 --np 4 -v 2

The following table is provided as a guide for several combinations and for several sets. In each case, the Pearson-R value is being reported for the set. As you can see, spending a little time to “tune” these settings for the various sets in your area(s) of interest can go a long way to providing you with improved results. Note: this table lists those structures which work well (e.g. CDK2, JNK1, p38, and so on), and those which work less well for rigid-docking with 25 MOE-provided poses (e.g. BACE and TYK2). Some additional tuning and including additional conformers/poses may take the improvement further.

MT-AMBER (Ensemble)
–nb-cutoff XX –ligand mol2 XX BACE CDK2 JNK1 MCL1 p38 PTP1B thrombin TYK2
13.0 13.0 0.30 0.74 0.62 0.18 0.59 0.72 0.44 0.18
20000.0 15.0 0.22 0.73 0.72 0.22 0.55 0.73 0.57 0.05
8.0 4.5 0.22 0.75 0.71 0.34 0.64 0.71 0.32 0.08
MT-AMBER (EndState)
–nb-cutoff XX –ligand mol2 XX BACE CDK2 JNK1 MCL1 p38 PTP1B thrombin TYK2
13.0 13.0 0.28 0.64 0.63 0.19 0.63 0.75 0.73 0.03
20000.0 15.0 0.22 0.73 0.72 0.21 0.55 0.73 0.57 0.04
8.0 4.5 0.29 0.26 0.63 0.30 0.62 0.75 0.53 0.01
MT-GARF (Ensemble)
–nb-cutoff XX –ligand mol2 XX BACE CDK2 JNK1 MCL1 p38 PTP1B thrombin TYK2
13.0 13.0 0.30 0.74 0.63 0.19 0.59 0.72 0.33 0.12
20000.0 15.0 0.22 0.73 0.72 0.22 0.55 0.74 0.53 0.05
8.0 4.5 0.23 0.75 0.71 0.36 0.63 0.70 0.24 0.06
MT-GARF (EndState)
–nb-cutoff XX –ligand mol2 XX BACE CDK2 JNK1 MCL1 p38 PTP1B thrombin TYK2
13.0 13.0 0.27 0.64 0.64 0.20 0.63 0.75 0.68 0.05
20000.0 15.0 0.22 0.73 0.72 0.22 0.55 0.74 0.53 0.05
8.0 4.5 0.29 0.67 0.65 0.32 0.65 0.77 0.47 0.03

 

Additional Examples

 

MTScoreE with MTDock

Please Note: MTDock is still undergoing significant refinement, optimization, and improvement. Until this initial effort is complete, it is recommended that you also run an external docker as noted in the previous tutorial. As noted in the previous tutorial, the MTScoreE (Ensemble) version of MTScore requires an ensemble of ligand poses to be generated and scored in order to sample the active site and accurately calculate a binding affinity. Generally, the better the poses the more accurate the affinity prediction, but the method can be quite forgiving. You can of course use your own docking package (e.g. MOE, GLIDE, etcetera) as each is different and each has its own strengths and weaknesses. However, MTScore can be used to score an ensemble of poses provided by its own built in Heatmap docking approach (MTDock). In this example, we demonstrate the use the MTDock algorithm found within qmechanic. In contrast to the use of an external docker, qmechanic can be run alone in the following manner: The input/output associated with this calculation is available in the Bace_030215_CAT_4p.tar.gz file.

${DIVCON_INSTALL}/bin/qmechanic Bace_030215.pdb --ligand Bace_030215_CAT_4p.mol2 --mtscore --np 4 -v 2
Command line: qmechanic Bace_030215.pdb Bace_030215_CAT_4p.mol2 --mtscore -v 2 --np 4 

EstimatedNumberOfConformers: 7776
1stLoop Scoring TIME: 0.00132575
2ndLoop Scoring TIME: 0.00138322
3rdLoop Scoring TIME: 0.0013916
FINAL SCORING TIME: 0.0188621
 Complexing series: 
Pocket Atoms 
/A/SER/71//N
/A/SER/71//O
.....
Ligand Atoms 
/L//1//O1 6
/L//1//CL1 5
....
20 count 249578 58916 62 126.628 secs
ligandCentroid:  14.9468   -1.467 -1.10704. pocketCentroid: 13.7522 -0.3792 -0.6573
TIME COMPUTE LOOP Placement: 0.501662
TIME BIG OUTER LOOP Placement: 34.771574
TIME totalSetupTime Placement: 0.000000
TIME totalTimePart1 Placement: 34.772811
--------------------------------------------------------------------------------
RB Optimization
Cycle Total Energy (kCal)       Delta E    Grad. Norm     Grad. Max   Coord. Norm
--------------------------------------------------------------------------------
    0     1.99542790e+07    1.9954e+07    1.7231e+09    1.1265e+09    0.0000e+00
    1     5.06428214e+06   -1.4890e+07    3.0058e+08    2.8245e+08    0.0000e+00
....
Protonating...
Discovered hydrogen network in 8.7394e-02 secs.
Computed hydrogen network penalties in 1.1204e-05 secs.
Protonation Complete.
Total Best Energy: 0.0000e+00 in 4.5696e-05 secs.
Protonating...
Discovered hydrogen network in 8.5995e-02 secs.
Computed hydrogen network penalties in 7.5250e-06 secs.
Protonation Complete.
Total Best Energy: 3.0971e+00 in 3.6435e-05 secs.
....
---------------------------------- Pose Score ----------------------------------
"Pose" "BF Intensity" "RMSD " "eSolv" "GARF Energy" "MTScoreES"
--------------------------------------------------------------------------------
0         -1.305      0.512     -9.712    -56.578     -8.604
1         -1.305      1.513    -10.603    -54.951     -7.425
4         -1.305      4.609    -14.785    -51.469     -6.974
2         -1.305      4.766    -13.549    -52.962     -6.920
3         -1.305      4.252    -11.499    -52.548     -6.435
--------------------------------------------------------------------------------

-------------------------------- Final MTScore ---------------------------------
"Species" "MTScoreES" "Binding ensembleZ" "Bound state" "Free state" "MTScoreE"
--------------------------------------------------------------------------------
"Bace_030215_CAT_4p"    -9.157335    -9.074305    -8.751572     0.000005    -8.751577
--------------------------------------------------------------------------------

Job Complete 
Total Computation Time (Seconds): 233.38592

The output is provided in tabular form with the following breakdown of information:

  • BFIntensiy = a value provided by MTCS which corresponds to the probability of the conformer.
  • RMSD = 3D structural root mean square deviation of the docked pose vs. the input mol2 pose.
  • MTScoreES = when in the Final MTScore table MTScoreES is the EndState binding affinity of the input mol2 file.
  • MTScoreE = when in the Final MTScore table, the MTScoreE is the Ensemble binding affinity of the ligand provided.

Additional Examples


MTScoreE with Molecular Dynamics (MD)

Docking is fast and with some additional tweaks (like increased sampling and additional optimization), docking coupled with MT can often yield accurate results. But what happens in those cases in which docking+MT doesn’t provide the results one expects or hopes? We know that sampling is dependent not only on the pose of the ligand within the active site, but also on how the protein changes upon binding. Most critically, it’s important to recognize that docking is really just a means to an end: global landscape minima generation for local (MT) sampling. Docking – which is “fast and dirty” is only one way to generate landscape minima. With recent improvements to the DivCon Discovery Suite (version DEV.746 and newer), you can also provide molecular dynamics (MD) generated snapshots both for the holo (protein:ligand complexed) structure and [optionally] for the apo structure as well. This method works very similarly to the method presented above for Induced Fit docking in that multiple protein:ligand complexes are provided to qmechanic for scoring. What is different is in how these conformers are generated. Whether you use AMBER, OpenMM, or a myriad of other approaches and implementations, most MD platforms have the ability to output bin’ed snapshots representative of the ensemble poses available (and in any case, numerous opensource tools exist to post-process trajectories to process and bin/filter trajectory data so regardless of the dynamics engine you use, you’ll likely have a tool to process it available).

  • First, review the documentation for your MD engine of choice. Since most of your time will be spent with your MD software, it is critical that you are familiar with how to run the software, what hardware may make the MD simulation quicker (GPU versus CPU), and how to generate production (e.g., post-equilibration) snapshots. Here are links to two popular packages.

AMBERMD: http://ambermd.org/tutorials/advanced/tutorial3/section2.htm

OpenMM: http://docs.openmm.org/latest/userguide/application.html?highlight=snapshot#a-first-example

  • Once you have generated a trajectory and the associated snapshots, you may use the PDB processing facilities available in –mtdock in order to provide them for the holo (protein:ligand complex) global landscape minima. In the example below, note the use of /L/LIG/123/ instead of a file after the –ligand command line argument. This is a URI style selection (e.g., /ChainID/RES/ResID/ format) within the complex.pdb file. You may of course use the standard “protein.pdb –ligand ligand.mol2” approach discussed in the tutorials above, but when working with trajectories, it’s easier simply select the ligand residue from the file and use that instead. DivCon takes care of making sure pockets are kept consistent from snapshot to snapshot and internal tests are performed to report whether there are problems with the selection provided. The complex_snaps_*.pdb option represents the snapshots. You may provide as many files as you like and you may name them whatever you wish. In this case, we assume the snapshots are named “complex_snap_1.pdb, complex_snap_2.pdb, complex_snap_3.pdb” and so on.

% qmechanic complex.pdb –ligand /L/LIG/123/ –mtdock complex_snap_*.pdb –mtscore ensemble -v 2

  • The command line above will only use the complex snapshots in the MT ensemble calculation. But there may be times that you also have a trajectory for the apo (or unbound) target. For the example below, we have expanded the above command line to include the –apo (unbound-protein) structures as well for the calculation of the ZP partition function. The software will test for congruence between the apo and holo structures so if the pocket residues don’t match, you should expect errors which are presented to help you along.

% qmechanic complex.pdb –ligand /L/LIG/123/ –apo apo_snap_*.pdb –mtdock complex_snap_*.pdb –mtscore ensemble -v 2

As a final note: these examples are provided based on MD-generated bound-protein and unbound-protein conformers. The identical logic applies if you wish to provide snapshots from other types of simulations (e.g., MD, MC, etcetera) or even experiments (e.g., X-ray, NMR, CryoEM, etcetera). The software itself has been built to address PDB files so it will work the same with whatever tool you use to generate the snapshots.

Additional Examples