Ligand Binding Pose Generation for FEP+

Tutorial Created with Software Release: 2025-4
Topics: Free Energy Perturbation (FEP), Hit-to-Lead & Lead Optimization, Small Molecule Drug Discovery
Products Used: FEP+, Glide

Tutorial files

20 MB
This tutorial is written for use with a 3-button mouse with a scroll wheel.
Words found in the Glossary of Terms are shown like this: Workspacethe 3D display area in the center of the main window, where molecular structures are displayed

 

Tip: You can hover over a glossary term to display its definition. You can click on an image to expand it in the page.
Abstract:

When ideating ligands during the lead optimization stage of a project, obtaining a plausible pose of each ligand within the target pocket is a prerequisite for running most physics-based calculations and simulations.
In this tutorial, you will learn how to use the FEP+ Pose Builder (Beta) to generate receptor-bound poses for congeneric or otherwise highly similar ligands for use in FEP+ and other workflows.

Tutorial Content

  1. Introduction

  1. Creating Projects and Importing Structures

  1. Prerequisites and System Preparation

  1. Generating Ligand Poses with the FEP+ Pose Builder

  1. Optional: Obtaining poses from MCS docking based ligand alignment

  1. Conclusion and References

  1. Glossary of Terms

1. Introduction

A good representation of how a ligand fits in a target pocket is essential for in-depth analysis and calculations to understand how a change in ligand structure affects the ligand-receptor interactions and stability of the complex. During the lead optimization stage, the initial lead(s) are modified by changing or adding individual functional groups and decorations to the ligand either by targeted ideation or enumeration strategies. This expansion around the initial lead results in a series of fairly similar ligands, which are used to probe the structure-activity relationship (SAR) of the ligand-receptor system.

For each of these ligand variations, it must then be determined how the modifications affect the binding affinity in order to identify promising ideas for further modification or synthesis. For any structure-based method for further analysis, in particular free energy perturbation (FEP), a prediction of the receptor-bound structure of each ligand in the series is required.

There is no universally optimal protocol for determining the best starting poses for FEP+, as differences in ligand chemistry and pocket geometry between systems bring distinct challenges. The main challenge in generating ligand poses for relative-binding FEP+ (RB-FEP+) calculations is in finding the right balance between maximizing overlap to the reference ligand and minimizing clashes with the receptor. Historically, finding the right pose required time and expertise, and used approaches like flexible ligand alignment, core-constrained docking, and manual adjustments.

Now, we recommend starting by using the new FEP+ Pose Builder (Beta), as it has outperformed other methods on benchmarks, and incorporates our best practice recommendations that would otherwise have to be done manually. See the Pose Generation for FEP documentation page for a comprehensive overview of alternative approaches.

For the purpose of this tutorial, you can imagine that you are working on a project to target BACE1 for treatment of Alzheimer’s disease. This fictional project is now in the lead optimization stage, with ongoing efforts focusing on three different ligand series. Your task is to set up and run FEP+ calculations on the new ligand ideas to guide design efforts.

You will use the FEP+ Pose Builder to generate FEP-ready poses for the BACE1 system that you can use in the BACE1 Inhibitor Design Using Free Energy Perturbation tutorial. In an optional second stage, you can use an alternative approach to obtain ligand poses using core-constrained docking with Glide via the MCS docking ligand alignment panel. Finally, you will compare the results from the FEP+ Pose Builder to those obtained by using core-constrained docking or flexible alignment to the reference ligands.

2. Creating Projects and Importing Structures

Figure 2-1. Importing the input data for this tutorial.

  1. Open Maestro and create a new project named FEP_pose.prj for this tutorial.
  2. Download the tutorial zip file including input files and reference outputs here: https://www.schrodinger.com/sites/default/files/s3/release/current/Tutorials/zip/fep_pose.zip
  3. After downloading the zip file, unzip the contents in your Working Directorythe location that files are saved for ease of access throughout the tutorial.
  4. Go to File > Import Structures.
  5. Find and choose FEP_pose_data.maegz from the tutorial files.

3. Prerequisites and system preparation

For this tutorial, you will use ligands from three different publications (you can find the full references at the end of this tutorial). Here is a brief overview of the ligand series and published data:

  • Series A (iminothiadiazinane warhead) and 5HTZ structure published by Scott et al. (Merck). Note that two ligand ideas (cpd98A and cpd99A) in this tutorial are not from this publication’s dataset, but were inspired by compounds 17 and 18 and included here for didactical purposes.
  • Series B (iminohydantoin warhead) and 4DJV structure published by Cumming et al. (Merck).
  • Series C (thiazine warhead) and 7DCZ structure published by Koriyama et al.  (Shionogi / Janssen).

Note that we are using pre-prepared receptor structures, which were generated using the Protein Preparation Workflow. Missing residues and loops were rebuilt using the automated workflow. All receptors were prepared with both catalytic aspartate residues deprotonated to keep the electrostatic environment in the pocket consistent between the different series. As none of the crystallographic waters in the structures are an integral part of the receptor, we have removed them from the structures for the alignment step. For more details on structure preparation, please see the Introduction to Structure Preparation and Visualization tutorial and the Best Practices for Protein Preparation. If you plan to run FEP+ calculations with this system, also consult the FEP+ Best Practices and Preparing Protein and Ligand Structures for FEP+.

4. Generating ligand poses with the FEP+ Pose Builder

4.1 Setting up the FEP+ Pose Builder workflow

  1. Find and open the FEP+ Pose Builder panel from the Tasks.

At every point where you provide data to the FEP+ Pose Builder (e.g. reference ligands or receptors, filtering criteria, or SMARTS patterns), you have the option to provide that data from within the Maestro Project (e.g. by loading included or selected entries) or load the configuration from a file. In this tutorial, you’ll mostly use the GUI as that makes it more transparent what the structures contain and how everything works. To see how the files (in particular, the SMARTS and receptor mapping files) should be structured or to reuse patterns/configuration made in the panel for another batch of ligands, check the output of an FEP+ Pose Builder job, for example the one included in the tutorial files.

Figure 4-2. Loading the ligand ideas.

The first step of the FEP+ Pose Builder workflow is to load in the ligand ideas.

  1. In the Specify Ligands section, click Browse to provide the ligand ideas from a file.
  2. Find and choose ligand_ideas.maegz from the tutorial files.
    • The panel updates to show that the file was successfully loaded (see the purple box in the figure)

Next, you can opt to prepare the ligands as part of the workflow. We recommend first validating that LigPrep/Epik gives robust results for your ligand chemistry by preparing your ligands outside the FEP+ Pose Builder. The option to generate tautomers should be used with caution – if you use it, make sure that the settings in the filtering stage and any SMARTS patterns used for alignment match the biologically relevant tautomer(s) for your system. You can further customize the LigPrep stage if needed by writing out the job files and editing the launch script to either pass command line arguments to LigPrep using -ligprep_args or provide a LigPrep input file as described here using -ligprep_file.

Figure 4-3. Enabling ligand preparation.

The default settings for ligand preparation work well for the dataset in this tutorial.

  1. Ensure the Prepare Ligands section is toggled on.

 

The workflow also includes an optional ligand filtering stage. In virtual screening applications, ligand filters are used to remove undesirable chemical moieties from the screening library, while here, the goal is to filter out particular ligand states generated by LigPrep/Epik that do not make sense for your particular pocket or series.

Figure 4-4. Configuring the ligand filtering step.

The BACE1 active site is strongly negatively charged, and all ligand series from our fictional project have a singly positively charged warhead. So we will apply a filter that only lets ligand states with a total charge of +1 through.

  1. Ensure that the Filter Prepared Ligands section is toggled on.
  2. Click Define…
  3. In the Ligand Filtering Panel, find and click Total_charge in the table in the General attributes tab.
  4. In the text field next to the == sign, write 1
  5. Click Add.
    • The Filter preview in the lower part of the panel (purple box in the figure) updates to add the new criterion.
  6. Click OK.

The next stage is the meat of the FEP+ Pose Builder workflow: Choosing the best reference for each of the ligands and aligning the ligand to this reference.

There are three different approaches to identify the best reference for the alignment for each ligand:

  • SMARTS or Similarity is the recommended default. It uses SMARTS patterns you specify to assign design ideas to reference ligands and falls back to your choice of either 2D similarity metrics (Bemis-Murcko or Maximum common substructure) or FEP+ 3D similarity if there is no appropriate match. 2D similarity is significantly faster than 3D similarity and sufficiently accurate in most cases. Assigning ligands to references by SMARTS is fastest.
  • Fixed reference uses the same reference ligand for all ideas. This mechanism could be helpful when aligning design ideas from an R-group enumeration or core hopping series back on to the single reference ligand.
  • Similarity only uses the similarity metrics as discussed above to choose the reference and make the alignment.

 

The reference structures must be prepared and have the correct binding mode in  the receptor that will be used for subsequent calculations. How many references you need will be project-specific: You will usually need at least one reference per congeneric series. If there are separate efforts to optimize particular parts of the ligand (e.g. a head group and tail group), it can be helpful to have multiple references for the underlying series. Also, if you plan to use cycle-closure RB-FEP+ with multiple reference compounds, consider including a reference ligand from each subseries to maximize similarity.

The different references do not need to bind the receptor in the same way or even in the same pocket. For the later clash resolution stage, you can assign the receptor that should be used for each ligand based on the reference to which it is matched.

Figure 4-5. Specifying reference ligands.

In this project, we’ll use four references. For our series B and C, a single reference has been sufficient to get good alignment for all ideas until now. For series A, there are two subseries differing in an aromatic linker (thiophene or phenyl), so we’ll use two references there.

  1. Optional: Includethe entry is represented in the Workspace, the circle in the In column is blue the reference ligands one by one or together to explore their shapes and functional groups.
  2. Select(1) the atoms are chosen in the Workspace. These atoms are referred to as "the selection" or "the atom selection". Workspace operations are performed on the selected atoms. (2) The entry is chosen in the Entry List (and Project Table) and the row for the entry is highlighted. Project operations are performed on all selected entries the reference_ligands group.
  3. In the Align Ligands section of the FEP+ Pose Builder panel, for Reference Ligands, choose Project Table (4 selected) and click Load.
    • The panel updates to confirm that the ligands have been loaded.
  4. Make sure the reference selection is set to SMARTS or Similarity.
    • A second option appears allowing you to switch between 2D and 3D similarity. Leave it at the default option (2D similarity).
  5. Click Define SMARTS…

We recommend defining a SMARTS pattern for each reference ligand to make sure ideas are aligned to the correct reference for their series. When choosing which atoms to include in the pattern, keep in mind that the pattern should be large and specific enough to only occur once per ligand. So pay attention when working with highly symmetrical ligands.

Remember that if a ligand fails to match the SMARTS pattern, the workflow falls back to 2D similarity metrics, so you’ll see this in the results and can adjust your pattern and re-run the workflow if needed.

Figure 4-6. Selecting the atoms forming the core for each of the reference ligands.

Repeat the following steps 16–18 for each reference ligand. See the image on the left for which atoms to select for each reference ligand.

  1. Include the reference ligand.
  2. Select the atoms forming the congeneric core in the workspace.

Note: Switching to Lasso selection from the main toolbar can help you quickly select the correct atoms.

There are multiple options to generate SMARTS for a structure: For example, you can right-click selected atoms in the 2D Sketcher and choose Copy As > SMARTS; you can also select atoms in the Workspace and use Edit > Copy As > SMARTS from Maestro’s main toolbar.

 

Within the FEP+ Pose Builder’s Define Smarts Rules for Reference panel, you can then double-click the Matching SMARTS cell for a given row to edit or paste a SMARTS string manually. Clicking the pencil icon in the cell opens the 2D sketcher, where you can directly draw the substructure from which to generate the SMARTS pattern.

Figure 4-7. Translating workspace selections to SMARTS rules for reference selection.

The FEP+ Pose Builder panel can directly load the workspace selection, shortcutting the process.

  1. In the Define Smarts Rules for Reference Selection panel, click Load from Selection.
    • A new row appears in the table with the SMARTS pattern and Reference populated.
  2. Repeat Steps 16-18 for each reference ligand. The image on the left shows how the panel should look fully populated.
  3. Click OK to return to the main workflow.

Note: You can also find the SMARTS strings for each of the references in the smarts_ref_input.csv file in the workshop archive and load them into the panel by using the Import Table… option.

Figure 4-8. Debugging SMARTS Patterns using the Workspace search bar.

Practical tip: It can be tricky to find robust SMARTS for your ligands, especially when the target substructures are aromatic or can have different tautomer states. To test or debug your patterns, you can use the search toolbar (opens with Ctrl/Cmd+F) to find atoms matching a SMARTS pattern in the workspace.

The final (optional, but strongly recommended) step of the FEP+ Pose Builder workflow is to resolve clashes of the aligned pose with a receptor structure. You can either provide a single receptor structure for all ligands or provide multiple receptors and specify a mapping between each reference ligand and receptor structure to use to resolve clashes.

We recommend removing all crystallographic waters from the receptors used here to generate ligand alignment. If you do not want to rely on GCMC to hydrate the binding pocket during your subsequent simulation, you can still use the generated ligand pose with a ‘wet’ receptor structure after alignment.

Figure 4-9. Comparing the three different receptor structures. 5HTZ ligand included to show the location of the binding site.

Optional: Include the three provided receptor structures with and without their corresponding ligands and compare differences in their binding pockets.

Figure 4-10. Specifying receptors for the clash resolution stage.

  1. Select(1) the atoms are chosen in the Workspace. These atoms are referred to as "the selection" or "the atom selection". Workspace operations are performed on the selected atoms. (2) The entry is chosen in the Entry List (and Project Table) and the row for the entry is highlighted. Project operations are performed on all selected entries the receptors group from the Entries.
  2. Confirm that the Resolve Clashes section in the FEP+ Pose Builder panel is toggled on.
  3. For Receptor, choose Project Table (3 selected).
  4. Click Load.
    • The panel updates to confirm that the receptors have been loaded.
  5. Click Map to Reference…

Figure 4-11. Mapping receptors to reference ligands.

You can now go through the table and set the appropriate receptor for each reference ligand. Both series A reference ligands can use the 5HTZ receptor structure.

  1. Click the Receptor column cell for each reference ligand in turn and pick the corresponding receptor. See the image on the left for how the panel looks fully populated.
  2. Click OK.

Figure 4-12. Launching the job.

Now, all that remains is giving your job a name and launching it.

  1. Change Job name to FepPoseBuilder_BACE1
  2. Click Run.
    • This job should take ~10 minutes. Feel free to use the pre-generated output files from the tutorial archive.

4.2 Analyzing and fine-tuning generated poses

Figure 4-13. Importing the pre-generated FEP+ Pose Builder results.

Optional: If you did not run the job yourself, import the pre-generated output files

  1. Go to File > Import Structures.
  2. Find and choose FepPoseBuilder_BACE1_out.maegz from the tutorial files.

Figure 4-14. Styling ligands in selected entries.

Optional: We recommend showing all ligand structures with the ball-and-stick representation for clarity.

  1. Select(1) the atoms are chosen in the Workspace. These atoms are referred to as "the selection" or "the atom selection". Workspace operations are performed on the selected atoms. (2) The entry is chosen in the Entry List (and Project Table) and the row for the entry is highlighted. Project operations are performed on all selected entries the FepPoseBuilder_BACE1 entry group heading.
  2. Near the top of the Entries pane, click the paintbrush icon to style atoms in the selected entries.
  3. Choose the Ligands option.
  4. Click the Ball-and-stick icon.

The output is structured as follows: At the top level, you’ll find a group for each receptor used to resolve clashes. Within each of these groups, you’ll find the receptor structure and a sub-group for each reference ligand mapped to this receptor. Each of these sub-groups in turn contains the structure of the reference ligand and up to two sub-groups containing the ligand ideas aligned to this reference, split by the method used to assign them to the reference (SMARTS or 2D Similarity).

 

To review the alignments, we recommend fixing each receptor and reference in the workspace and stepping through the ligands aligned to that reference. Let’s start with series B, where the analysis is straightforward as all ligand ideas there successfully matched the SMARTS pattern.

Figure 4-15. Visualizing the aligned poses for the series B ligands (green carbons) together with the reference (white).

  1. Expand and select the 4DJV - prepared - dry (series B) group.
  2. Double-click the In circles to fix the 4DJV - prepared - dry receptor and 4DJV_ligand entries in the workspace.
  3. Use the left and right arrow keys to step through the poses to identify poses potentially needing manual refinement.

When analyzing the ligand poses, there are no hard and fast rules for judging which poses are correct or optimal. Keep in mind that there is usually a significant difference between a good docked pose and a good FEP-ready pose: The main purpose of a pose from docking is to provide a realistic impression of whether and how the ligand would bind in the pocket, maximizing interactions and minimizing clashes. Poses to be used as FEP+ inputs should maximize atom overlap to the reference to improve prediction accuracy, even if that causes bad clashes with the receptor which resolve with minor reorganization during the MD equilibration stage.

In the limit of perfect (infinite) sampling during the MD simulation, the input geometry of the ligand becomes irrelevant. However, starting from a “bad” pose in an FEP+ calculation can trap the design idea in an unrealistic configuration in the binding pocket or require excessive computational time for the necessary ligand and/or receptor reorganization, which would significantly reduce the accuracy of the binding affinity predictions.

The main thing to look out for when evaluating ligand poses besides atom overlap with the reference is therefore unresolvable clashing with the receptor (i.e., clashes that couldn’t be resolved with minor reorganization during MD equilibration), mainly “ring spears” like the lysine sidechain threading through the phenyl ring in this image:

Figure 4-16. A Lysine residue passing through a ligand ring (“ring spear”). This type of clash cannot be readily resolved during MD equilibration.

Less problematic are unfavorable torsions, e.g. ecliptic conformations, and other steric and functional clashes with the receptor, e.g. two H-Bond acceptors oriented directly towards each other. These unfavorable interactions are usually relaxed readily within the MD equilibration.

 

 

There are a few ligand ideas from this series where the o-substituent on the final phenyl is pointing in the wrong direction. While the rotation of the phenyl ring is usually sampled within the timescale of the FEP+ simulation (and using “enhanced dihedral sampling” in the FEP+ simulations often encourages this sampling), in confined regions of the binding pocket, it can be more challenging to reliably interconvert between both phenyl orientations. Thus, with your experience and understanding of the binding pocket, you might identify when it would be advantageous to manually rotate the appropriate dihedral to adopt the optimal initial FEP+ pose. In a real project, we would recommend introducing additional references for this ligand series, to allow the FEP+ Pose Builder to automatically match the orientations for the substituents of the terminal phenyl or correctly align the bulkier substituents at the chiral center.

Figure 4-17. Locking the reference ligand structure to prevent accidental edits.

First, make the reference ligand non-editable so you can keep it in the workspace as a visual reference for alignment without accidentally modifying it.

  1. Right-click In circle for the 4DJV_ligand (series B) entry and choose Lock (set non-editable).

Figure 4-18. Entering Rotate Dihedral mode.

Next, repeat the following steps for each ligand where the substituent is misaligned (CAT-4n, CAT-4o, CAT-24, CAT-17i, CAT-13a, CAT-17f, CAT-17g, CAT-17e)

  1. Include the ligand entry.
  2. Right-click the bond between the two phenyl rings and choose Rotate Dihedrals.

Figure 4-19. Rotating the phenyl substituent.

  1. Left-click and drag to the left or right to rotate the group until the substituents are aligned.
  2. Click OK in the banner to confirm the edit.
  3. Repeat the last four steps for the next ligand.

 

Note: You can undo confirmed edits by pressing Ctrl/Cmd+Z.

The ligand poses from this series are now nicely aligned and ready for FEP+ simulations. In a real project, we would recommend adding additional references for the different subseries using this warhead so that the bulkier substituents at the chiral center can be aligned even better through the workflow itself.

If you want to learn how to run RB-FEP+ calculations on this ligand series, you can continue with the BACE1 Inhibitor Design Using Free Energy Perturbation tutorial. For additional examples and ideas on troubleshooting FEP+ Pose Builder outputs, read on here.

4.3 Troubleshooting misaligned poses

This section will use the other two ligand series in the data set to highlight some stumbling blocks you might encounter when using the FEP+ Pose Builder in your project. Feel free to explore and experiment and see if you can obtain better results than from our initial configuration.

As a general troubleshooting tip, it might be helpful to check the FEP+ Pose Builder job files, where you can find the intermediate results and logs from each stage of the workflow. This is particularly useful if you observe ligand ideas ‘going missing’: If you don’t get an output pose for a ligand idea, there might have been an issue with the input structure, during preparation, or the ligand may have failed to pass the filtering stage. Check the log files for the corresponding stages in the ligprep or ligfilter subdirectories of your job folder to find out what happened.

You can now move on to series A, where all 9 ligands correctly matched their corresponding reference — the 5HU1 ligand for those with a phenyl linker and the 5HTZ ligand for those with a thiophene.

Figure 4-20. Inspecting results for series A. Reference ligand in blue, aligned pose in green.

  1. Right-click a blank spot on the workspace and choose Clear Workspace.
  2. Expand and select(1) the atoms are chosen in the Workspace. These atoms are referred to as "the selection" or "the atom selection". Workspace operations are performed on the selected atoms. (2) The entry is chosen in the Entry List (and Project Table) and the row for the entry is highlighted. Project operations are performed on all selected entries the 5HTZ - prepared - dry (series A) group.
  3. Double-click the In circles to fix the 5HTZ - prepared - dry receptor and 5HU1_ligand entries in the workspace.
  4. Use the left and right arrow keys to step through the poses to identify poses potentially needing manual refinement.

Figure 4-21. Inspecting results for the thiophene subseries of series A.

Repeat the previous steps with the 5HTZ_ligand fixed in the workspace instead of the 5HU1_ligand to step through the results for the subseries with the thiophene linker.

The poses for all series A ligands aligned to the 5HU1 ligand look good to use without further adjustment. However, the tails for the ligands with the thiophene linker (cpd98A and cpd99A) are not well-aligned to the 5HTZ reference.

In a real project, how to resolve this will depend on where the project team is headed: If you expect more design ideas to include the thiophene linker in the future, you’ll want to go back and revise the SMARTS pattern so these ligands can be handled by the FEP+ Pose Builder. If only the other series/scaffolds are actively explored, you could instead obtain the alignments for those two ligand ideas with e.g. MCS docking ligand alignment or manual alignment.

 

Additionally, two ligands from series C, cpd12C and cpd14C, failed to match the series C SMARTS pattern and were aligned to the 5HU1 reference ligand for series A based on 2D Similarity. Peeking for a second into the results for series C, you’ll see that cpd13C_2 also failed to match the SMARTS pattern, but its 2D similarity to the 7DCZ reference ligand was higher than to the 5HU1 ligand. You can now investigate these cases further to understand why this has happened.

Figure 4-22. Investigating ligands aligned to the wrong reference.

  1. Right-click a blank spot on the workspace and choose Clear Workspace.
  2. From the 5HTZ - prepared - dry (series A) group, includethe entry is represented in the Workspace, the circle in the In column is blue the 5HU1_ligand (series A), cpd12C, and cpd14C.
  3. Additionally, includethe entry is represented in the Workspace, the circle in the In column is blue the 7DCZ_ligand (series C), and cpd13C_2 from the 7DCZ - prepared - dry (series C) group.
  4. Press Ctrl+L/Cmd+L to tile the workspace and compare the structures.

It seems like the cpd12C, cpd13C_2, and cpd14C ligand ideas have a C-C single bond rather than a double bond in their head group. This causes the SMARTS pattern to fail to match the reference, and the algorithm falls back to use 2D similarity. The change from a double bond to a single bond introduces significant non-planarity to the head group, so you might need to align one of the affected ligands with an alternative method and use it as an additional reference for the others.

Figure 4-23. Inspecting results for series C. Reference ligand in pink, aligned pose in green.

Optional: Repeat the analysis with the series C results.

  1. Right-click a blank spot on the workspace and choose Clear Workspace.
  2. Expand and select the 7DCZ - prepared - dry (series C) group.
  3. Double-click the In circles to fix the 7DCZ - prepared - dry receptor and 7DCZ_ligand entries in the workspace.
  4. Use the left and right arrow keys to step through the poses to identify poses potentially needing manual refinement.

All series C ligand ideas that matched the reference SMARTS pattern produced nicely aligned poses without need for further adjustments.

5. Optional: Obtaining poses from MCS docking based ligand alignment

As an alternative approach for generating aligned ligand poses, you can now try optionally using MCS docking based ligand alignment on the series C ligands. The shared core of the ligands within the series allows us to use core constrained (also known as maximum common substructure (MCS) based) docking, using the known pose of the reference ligand. This constraint ensures that a similar binding mode is maintained for all ligands, while still using the conformer generation and scoring protocols in Glide.

The MCS docking ligand alignment panel provides a streamlined interface to the Glide backends for grid generation and docking for the purpose of ligand alignment to one or multiple reference ligands. Should docking fail to produce a pose for a given ligand, flexible ligand alignment without constraints is used as a fallback. However, note that this approach to obtaining alignments is significantly slower than the FEP+ Pose Builder. In particular for large data sets, we do not recommend using MCS docking ligand alignment as the main pose generation approach.

For a detailed introduction to using Glide for docking of small-molecule ligands, see the Structure-Based Virtual Screening Using Glide tutorial. Note that Glide considers anything included in the structure that is not explicitly flagged as ligand to be a rigid part of the receptor. This includes any crystallographic water molecules present in the structure.

5.1 Setting up the MCS Docking Ligand Alignment

Figure 5-1. Opening the MCS Docking Ligand Alignment panel.

First, include the receptor and reference ligand for series C in the Workspace and open the MCS docking ligand alignment panel.

 

  1. Include the series C receptor (7DCZ - prepared - dry) and reference ligand (7DCZ_ligand) entries.
  2. Go to Tasks > Browse > Structure Alignment > MCS Docking Ligand Alignment
    • The MCS Docking ligand alignment panel opens.

Figure 5-2. The Receptor section of the MCS docking ligand alignment panel.

Next, generate a receptor grid to use during docking. The cocrystal ligand is used to define the location of the binding pocket for the receptor. 

 

  1. For Receptor Grid, choose Create from Workspace and click Set Up Grid

Figure 5-3. Setting up a Grid from the MCS docking ligand alignment panel with the reference ligand picked in the workspace.

  1. In the MCS Create Receptor Grid panel, click Pick to Identify Ligand
  2. Left-click a ligand atom in the workspace.
  3. Change the Grid file name to 7DCZ_grid.zip and click OK.
    • This job takes about a minute. The resulting grid file is not incorporated into the workspace but saved to the working directory.  

Note: It is also possible to create the grid using the Receptor Grid Generation panel, which allows changing additional options.

Next, you will need to specify the ligand ideas to dock. Note that in contrast to the FEP+ Pose Builder, you must provide prepared ligand structures to the MCS Docking Ligand Alignment panel. When working with ligands for your own project, keep in mind that LigPrep may output multiple viable states for a given ligand. As usual with Glide, only the protomers or tautomers explicitly included in the ligand file will be aligned.

Figure 5-4. Specifying Ligands for alignment and the reference ligand.

You can find the prepared and filtered series C ligands in the tutorial files.

  1. In the MCS Docking Ligand Alignment Panel, in the Ligands for Alignment section click Browse.
  2. Find and choose seriesC_ligands_prepped.maegz in the tutorial files.
  3. Select(1) the atoms are chosen in the Workspace. These atoms are referred to as "the selection" or "the atom selection". Workspace operations are performed on the selected atoms. (2) The entry is chosen in the Entry List (and Project Table) and the row for the entry is highlighted. Project operations are performed on all selected entries the 7DCZ_ligand (series C) entry.
  4. In the Reference Ligands section, choose Project Table (selected entries)

When applying this to your own data set, additional options in this panel can be helpful:

 

  • Similarly to the FEP+ Pose Builder, the MCS docking ligand alignment panel allows you to use multiple reference ligands and receptors. You need to generate a receptor grid for each receptor you plan to use, and can specify the mapping between reference ligands and receptor in the Advanced Options.
  • As mentioned above, if the core-constrained docking stage does not return a pose for a ligand idea, the panel falls back to flexible ligand alignment (using the backend of the Ligand Alignment panel). You can increase the likelihood of obtaining a docked pose by allowing docked ligands to become reference ligands for docking other ideas.

Figure 5-5. Running the MCS docking ligand alignment job.

By default, the results from this job are not automatically incorporated into your project.

  1. Click the cogwheel symbol to open the Job Settings.
  2. Change the Incorporate mode to Append new entries as a new group.
  3. Change the Name to mcs_docking_seriesC and click Run.
    • This job should take approximately 10 minutes.

5.2 Analyzing poses from MCS docking based ligand alignment

In this section, you will inspect the poses for the series C ideas obtained from the MCS docking ligand alignment, and compare them to the poses from the FEP+ Pose Builder for this series.

Figure 5-6. View poses of the ligand ideas obtained from MCS docking ligand alignment.

Optional: If you did not run the job yourself, import the pre-generated output files.

  1. Go to File > Import Structures
  2. Find and choose mcs_docking_seriesC_out_pv.maegz from the tutorial files.

Figure 5-7. Adjust visualization settings to easily compare poses.

Optional: You may wish to include the whole mcs_docking_seriesC_out_pv group and double click the Presets button or choose Binding Mode Comparison in the Presets menu to have each ligand represented in a unique color. Additionally, you can turn on interactions using the workspace toggle in the lower right corner of the screen to visualize clashes.

Figure 5-8. Stepping through the MCS docking ligand alignment results for the series C ideas.

  1. Double-click the In circles to fix the 7DCZ - prepared - dry receptor and 7DCZ_ligand entries in the workspace.
  2. Select(1) the atoms are chosen in the Workspace. These atoms are referred to as "the selection" or "the atom selection". Workspace operations are performed on the selected atoms. (2) The entry is chosen in the Entry List (and Project Table) and the row for the entry is highlighted. Project operations are performed on all selected entries the mcs_docking_seriesC_out_pv group
  3. Use the left and right arrow keys to step through the entries one by one and inspect the ligand poses.

The first five entries of the output group are the ligand ideas where a docked pose could not be obtained, so those poses were obtained using flexible ligand alignment. The ligands listed below the receptor entry (cpd13C_2 and cpd18C_1) have poses obtained from core-constrained docking.

Figure 5-9. Comparing poses for the cat12C (left) and cat14C (right) ligands to the reference ligand (pink).

First, compare the new poses for the cat12C and cat14C ligands to those from the FEP+ Pose Builder.

  1. Fix the 7DCZ_ligand entry in the workspace.
  2. Includethe entry is represented in the Workspace, the circle in the In column is blue both the cpd12C and cpd14C entries from the mcs_docking_seriesC group and the cpd12C and cpd14C entries from the FepPoseBuilder_BACE1 output.
  3. Press Ctrl+L/Cmd+L to view the alignments side-by-side.

 

Note: The order of the tiles depends on the order of inclusion into the workspace. To identify which tile corresponds to which entry, hover over the structure in one of the tiles and check which entry is highlighted. You can also click the workspace properties in the top-leftmost tile and add the Source File property to see where each structure came from at a glance.

While core-constrained docking did not return poses for those ligands, the fallback poses from flexible ligand alignment are very well aligned to the reference ligand. Although the head group is forced into a planar conformation by the core constraints, this strain should relax during an MD or FEP+ simulation. In our experience, prioritizing the overlap of unperturbed atoms over ligand strain and protein interactions results in more robust atom mapping and less noisy FEP+ results.

Figure 5-10. Comparing poses for the cat13C_2 ligand to the reference ligand (pink).

Next, compare the new (docked) pose for the cpd13C_2 ligand to the FEP+ Pose Builder pose.

  1. Repeat step 7 for the corresponding cpd13C_2 ligand entries.

This ligand pose was obtained from core-constrained docking. Due to the differences in the headgroup, the alignment favors the tail of the ligand. Neither of the two poses is very good for this ligand idea, as both the FEP+ Pose Builder and MCS docking ligand alignment struggle to position the head group. For this ligand, you may need to align the pose manually or using the Ligand Alignment panel.

Figure 5-11. Comparing poses for the cat18C ligand to the reference ligand (pink).

Finally, compare the new (docked) pose for the cpd18C_1 ligand to the FEP+ Pose Builder pose.

  1. Repeat step 7 for the corresponding cpd18C_1 ligand entries.

For this ligand idea, the pose from the FEP+ Pose Builder shows excellent alignment to the reference in both the head group matched by the SMARTS pattern and the rest of the ligand. The pose from core-constrained docking is not as well aligned, showing significant deviations in the head group and smaller deviations in the rest of the ligand.

5. Conclusion and References

In this tutorial, you used the FEP+ Pose Builder to predict aligned poses for ligand ideas from three series (designated A, B, and C for short) for use in FEP+ calculations. You set up the FEP+ Pose Builder calculation, including ligand preparation and filtering as well as clash resolution. After running the calculation, you inspected the aligned poses and made manual adjustments to the alignment of ligands from series B based on your system knowledge. For the other two series, you investigated cases where the FEP+ Pose Builder did not produce satisfactory poses. Finally, you used the MCS Docking Ligand Alignment panel to generate alternative poses for the ligands from series C and compared those results to those obtained from the FEP+ Pose Builder.

In summary, while there is no generic workflow to obtain optimal poses, the FEP+ Pose Builder streamlines and automates the process of obtaining good poses for larger batches of congeneric ligands. In cases where the FEP+ Pose Builder struggles, using the MCS docking ligand alignment panel to perform a combination of core-constrained docking and flexible ligand alignment can give alternative poses. Poses obtained from either one of the methods can then additionally be manually refined if needed to achieve a high level of accuracy for the following calculations.

You can find an end-to-end example of setting up, running and evaluating an RB-FEP+ calculation for the series B ligands used in this tutorial in the BACE1 Inhibitor Design Using Free Energy Perturbation tutorial.

For further reading:

 

 

6. Glossary of Terms

Entry List - a simplified view of the Project Table that allows you to perform basic operations such as selection and inclusion

included - the entry is represented in the Workspace, the circle in the In column is blue

pose - conformation of a ligand within a pocket e.g. as determined by docking

Project Table - displays the contents of a project and is also an interface for performing operations on selected entries, viewing properties, and organizing structures and data

Receptor grid - a computationally efficient representation of a receptor for docking purposes

Scratch Project - a temporary project in which work is not saved, closing a scratch project removes all current work and begins a new scratch project

selected - (1) the atoms are chosen in the Workspace. These atoms are referred to as "the selection" or "the atom selection". Workspace operations are performed on the selected atoms. (2) The entry is chosen in the Entry List (and Project Table) and the row for the entry is highlighted. Project operations are performed on all selected entries

Working Directory - the location that files are saved

Workspace - the 3D display area in the center of the main window, where molecular structures are displayed