Active Learning FEP+ (Beta) Panel

Screen a large ligand library with FEP+, using active learning to prioritize and identify the most promising compounds.

  • Using
  • Features
  • Additional Resources

Using the Active Learning FEP+ (Beta) Panel

Active Learning FEP+ screens a large ligand library to identify the most promising compounds, using machine learning to decide which ligands to evaluate with FEP+ in each iteration. This approach explores a large chemical space with far fewer FEP+ calculations than screening the entire library.

Choose the calculation mode with the Relative and Absolute buttons. In Relative mode, the library ligands are ranked by relative binding free energy against one or two reference ligands; in Absolute mode, they are ranked by absolute binding free energy. The inputs and settings available in the panel differ between the two modes.

Prepare your inputs before you start a job. In Relative mode, the ligand library must be aligned with the reference ligands; in Absolute mode, the ligand library must be docked to the receptor. A warning is displayed if you start a job with inputs that are not aligned or docked.

In each iteration, the specified number of ligands is evaluated with FEP+, and the machine learning model is updated and used to select the ligands for the next iteration. Set the number of iterations, the number of ligands per iteration, and the selection rule in the Settings section.

To write out the input file and a script for running the job from the command line, click the arrow next to the Settings button and choose Write. For information on command usage and options, see fep_active_learning.py Command Help for a Relative calculation, or ab_fep_active_learning.py Command Help for an Absolute calculation.

Active Learning FEP+ (Beta) Panel Features

Relative / Absolute buttons

Select the calculation mode. In Relative mode, the panel calculates relative binding free energies against one or two reference ligands; in Absolute mode, it calculates absolute binding free energies. The inputs and settings available in the panel differ between the two modes.

Inputs section

Specify the receptor, ligand library, and reference ligands for the calculation. All inputs must be prepared before you start the job. In Relative mode, the ligand library must be aligned with the reference ligands; in Absolute mode, the ligand library must be docked to the receptor. A warning is displayed if you start a job with inputs that are not aligned or docked.

Receptor option menu

Use the menu and the Load/Browse button to specify the receptor structure, from entries selected in the Workspace or from a file in Maestro format (.mae, .maegz). The receptor input must contain exactly one receptor, with no non-receptor components other than water or membrane structures.

Ligand library option menu

Use the menu and the Browse button to specify the library of ligands to screen, from a file in SMILES or Maestro format (.smi, .mae, .maegz). The file must contain at least 50 ligands, and the structures cannot contain any non-ligand components.

Reference ligand(s) option menu

Use the menu and the Load/Browse button to specify one or two reference ligands, from entries selected in the Workspace or from a file in Maestro format (.mae, .maegz). The structures cannot contain any non-ligand components. If you specify two reference ligands, they must have unique titles. This input appears in Relative mode only.

Specify experimental affinity button

Add binding affinity data to the reference ligands. The data must be available in the structure source. Click Define to open a pane, where you can choose the property and set the affinity units. Once you have made the selection, the property and its units are displayed in the panel. This input appears in Relative mode only.

Settings section

Configure the active learning and FEP+ settings for the calculation.

FEP+ Advanced Options link

Open the FEP+ Advanced Options Dialog Box, where you can set simulation parameters, sampling, system builder, and force field options for the underlying FEP+ calculations. Simulation parameters and sampling are set separately for the Single Edge Map and Final Full Cycle Closure stages (labeled Absolute Binding and Final Rescore Absolute Binding in Absolute mode). Missing force field parameters are generated automatically. The sections available depend on whether you selected Relative or Absolute mode.

Optimization Target option menu

Select the property to optimize during active learning:

  • ΔG: Binding Free Energy Change—improve binding potency.
  • LE: Ligand Efficiency—improve binding potency per heavy atom.
  • LLE: Lipophilic Ligand Efficiency—balance potency and lipophilicity.
  • AEI: ADME Efficiency Index—balance potency with absorption, distribution, metabolism, and excretion.
Run Glide option

Dock the ligand library with Glide to generate features before FEP+ scoring. This option appears in Absolute mode only.

Optimize For option

Select how the Glide results are used: Target Value to optimize toward the optimization target, or Diversity to maximize chemical diversity.

Grid file button

Click Browse to select the Glide grid file to use for docking the library.

Number of iterations text box

Set the number of active learning iterations to run, from 1 to 7. The default is 3.

Customize iterations option

Customize the number of ligands and the selection rule for individual iterations. Click Define to set the per-iteration values.

Ligands per iteration text box

Set the number of ligands to evaluate with FEP+ in each iteration, from 50 to 999. The default is 250.

Selection rule option menu

Select the rule used to choose which ligands to evaluate in the next iteration: Greedy, Most Uncertain, Random, DISE, or Diversity.

Use custom core option

Define a custom core for the single edge map that is used to map the library ligands. Click Define to specify the core as a SMARTS pattern, which you can type directly or derive from atoms selected in the Workspace. For more information on using a custom core, see Using a Custom Core in the FEP+ Panel. This option appears in Relative mode only.

Final full cycle closure option

Select Skip to omit the final full cycle closure stage. In Absolute mode, this section is labeled Final rescore absolute binding (and the Single edge map section is labeled Absolute binding). When you skip this stage, its settings are also hidden in the FEP+ Advanced Options Dialog Box.

If final full cycle closure is skipped and two reference ligands are loaded, only the first reference ligand is used for the calculation. In Relative mode, this setting is disabled if only one reference ligand is loaded.

Number of best ligands text box

Set the number of top-ranked ligands to carry into the final full cycle closure, from 1 to 999.

Submit to Web Services option

Run the FEP+ calculations through FEP+ Web Services rather than on a local host. This option is available only if you have purchased FEP+ Web Services.

Job toolbar

Manage job submission and settings. See Job Toolbar for a description of this toolbar.

The Job Settings button opens the Active Learning FEP+ - Job Settings Dialog Box, where you can make settings for running the job.

Status bar

Use the Reset button to reset the panel to its default settings and clear any data from the panel. If the panel has a Job toolbar, you can also reset the panel from the Settings button menu.

If you can submit a job from the panel, the status bar displays information about the current job settings and status for the panel. The settings include the job name, task name and task settings (if any), number of subjobs (if any) and the host name and job incorporation setting. The job status can include messages about job start, job completion and incorporation.

The status bar also contains the Help button , which opens an option menu with choices to open the help topic for the panel (Documentation), launch Maestro Assistant, or if available, choose from an option menu of Tutorials. If the panel is used by one or more tutorials, hover over the Tutorials option to display a list of tutorials. Choosing a tutorial opens the tutorial topic.

Related Topics