AutoDesigner - R-Group Design Panel

Set up and run an AutoDesigner workflow to explore R-group chemical space with docking-based scoring and filtering.

To display this panel, do one of the following:
  • click the Tasks button and browse to De Novo Design → AutoDesigner R-Group Design
  • Overview
  • Using
  • Features
  • Additional Resources

Overview

AutoDesigner R-Group Design is a de novo design workflow that explores R-group chemical space to generate novel compounds with docking-based scoring and filtering. The workflow keeps the common core fixed while replacing the R-group at a specified position. AutoDesigner routinely explores on the order of 100 million to over 1 billion molecules, with typical runtimes of 24 to 48 hours. The generated compounds can be used as input for Active-Learning FEP+.

AutoDesigner R-Group Design is best suited for hit-to-lead and lead optimization stages of drug discovery, where the goal is to optimize a single R-group position while keeping the core scaffold fixed. It requires a high-quality protein structure and a validated FEP+ model. AutoDesigner is target-agnostic and can be applied to any structurally enabled target.

The algorithm uses four methods to explore chemical space:

  1. Matched molecular pair (MMP) transforms — applies over 300 million transforms derived from PubChem and ChEMBL databases to generate new R-group variations. Before applying transforms, a set of input compounds is generated through recursive trimming of the starting ligand followed by R-group decoration and clustering, to maximize the number of applicable transformations.

  2. PathFinder reaction-based enumeration — performs exhaustive retrosynthetic analysis of the input molecules, followed by combinatorial synthesis of the resulting reaction pathways using commercially available reactant libraries to explore synthetically accessible R-groups. The reactant containing the common core is kept constant while the other reactants are varied.

  3. Recursive trimming — breaks predefined single bonds in a recursive fashion while keeping the bonds of the immutable core intact, generating a large number of additional ligands from all possible relevant permutations. Rules are applied to prevent trimming of bonds that would lead to undesirable outputs (e.g., aromatic bonds).

  4. R-group decoration — decorates compounds with curated R-group libraries at every available site excluding the fixed core. Two rounds of decoration ensure broad exploration of the available chemical space.

AutoDesigner uses progressive filtering cascades. Intermediary filtering stages use fewer filters and allow compounds through that could potentially be rescued in later stages, for example, compounds are not filtered based on the number of hydrogen bond donors as these can be added or removed in the decoration stage. The final filtering cascade uses all available filters with the strictest settings to retain only compounds meeting all criteria. The filtering cascade includes over 1,500 proprietary SMARTS patterns that capture basic chemistry rules, chemical stability, and medicinal chemistry liabilities. These patterns are continuously updated based on medicinal chemistry feedback. The substructure-based filtering is automatically adjusted each run to ensure the starting ligand and common core pass the substructure filter.

Using the AutoDesigner - R-Group Design Panel

To set up an AutoDesigner job:

  1. Prepare the Glide input files: a grid file, a docking input file with core constraints, and a reference ligand file. For detailed instructions, see AutoDesigner R-Group Design Documentation .

  2. In the Setup tab, load a starting ligand and click Edit in Sketcher to define the common core. Use the R+ button to indicate the R-group position to explore, then click Save.

  3. Configure LigPrep and Glide Docking settings, or use the defaults. Select the docking input file, reference ligand file, and grid file.

  4. In the Filters tab, configure the Target Property Space, Excluded SMARTS, FEP Amenability Filter, and any additional filters as needed.

  5. Configure job settings. You should use a minimum of 1,024 processors. The driver node should have 8 cores, 64 GB RAM, and 2 TB SSD scratch. Compute nodes should have 16 cores, 32 GB RAM, and 200 GB scratch.

  6. Click Run to submit the job, or use the Write option to save job files for manual submission on a cluster.

The job produces the full set of output structures (autodesigner_full.maegz), a randomly selected sample of 1,000 compounds (autodesigner.maegz), a PDF report (autodesigner_report.pdf), and a log file (autodesigner.log).

You should choose a starting ligand with an existing R-group at the enumeration position (R≠H). This ensures the property space aligns with ligands that already have an R-group, the SMARTS filters are adjusted correctly, the PathFinder enumeration finds applicable routes, and the MMP stage produces more outputs. For additional guidance, see AutoDesigner Best Practices.

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 autodesigner.py Command Help.

AutoDesigner - R-Group Design Panel Features

The panel is organized into two tabs: Setup and Filters.

Setup tab

Filters tab

Setup tab

Starting Ligand text and load button

Select a single ligand structure as the starting input for the (Undefined variable: Snippet.product_name) algorithm. Import from the Project Table or Workspace and click Load. The starting ligand is neutralized upon loading. Only neutral, single-ligand structures are accepted.

Common Core display area

Displays a 2D representation of the common core defined for the starting ligand. After loading a starting ligand, a copy of the ligand appears here. Use Edit in Sketcher to mark the R-group position and define the common core.

Edit in Sketcher button

Click this button to open the 2D sketcher dialog. Use the R+ button to place an attachment point indicating the R-group position to explore, then click Save. The common core (everything except the marked R-group) is kept fixed during the workflow. The sketcher only accepts neutral structures. The fixed common core cannot contain isotope-labeled atoms.

Use custom settings option

Select this option to enter custom LigPrep settings in the text box. If not selected, the default (Undefined variable: Snippet.product_name) LigPrep settings are used, which generate tautomers and ionization states with Epik, produce up to 32 stereoisomers per input structure, and apply a pH tolerance of 0.5.

Use custom filter file option

Select this option and click Browse to upload a custom LigPrep property filter file. If not selected, standard (Undefined variable: Snippet.product_name) filter files are generated and used. The default filter removes ligands with an Epik state penalty greater than 1 in intermediary filtering cascades and greater than 0.5 in the final filtering cascade.

Docking input file file path and Browse button

Click Browse to select a Glide input file (.in). The docking model must include core constraints. The intermediary filtering cascades use only the core constraint, while the final filtering cascade uses all constraints present in the input file, including any hydrogen bond or positional constraints.

Reference ligand file file path and Browse button

Click Browse to select a reference ligand file (.reflig.maegz) that defines the reference structure and core constraint for docking. It is recommended to use the fixed common core as the reference ligand and the All heavy atoms setting for core atoms to ensure aligned output poses.

Grid file (zip) file path and Browse button

Click Browse to select a Glide grid file (.zip) for the target receptor.

AutoDesigner Best Practices link

Click this link to display the AutoDesigner Best Practices topic, which provides guidance on starting ligand selection, LigPrep charge state validation, and docking model setup.

Define Filters link

Click this link to switch to the Filters tab. The label changes to Change Filters after any filter settings have been modified.

Filters tab

Excluded SMARTS file path and Browse button

Click Browse to select an optional SMARTS file (.sma) containing patterns to exclude from the (Undefined variable: Snippet.product_name) output. Each line in the file specifies one SMARTS pattern, and compounds matching any pattern are filtered out. Use this to exclude previously explored series or competitor compounds.

Target Property Space Select button menu

Click Select to configure or import Target Property Space settings. Choose Configure to open the Target Property Space dialog, where 21 physicochemical properties can be set with minimum and maximum values. The property ranges are automatically populated from the descriptors calculated for the starting ligand, and they update whenever the starting ligand is changed. Choose Import to load previously saved settings from a .propspace file. By default, the property space must include the starting ligand properties. Uncheck Require property space include starting ligand to allow the property space to fall outside the starting ligand values. Settings can be saved using the Export button in the dialog.

The 21 configurable properties include molecular weight, fraction sp3, PSA, AlogP, total rotatable bonds, consecutive rotatable bonds, chiral centers, ring counts (total, aromatic, saturated), spiro atoms, bridgehead atoms, hydrogen bond acceptors, hydrogen bond donors, maximum ring size, sulfur atoms, alkynes, quaternary carbons, 5,5-fused ring systems, ligand total charge, and ligand absolute total charge.

FEP Amenability Filter section

Filters compounds to retain only those suitable for relative binding free energy calculations. A compound is considered amenable if the number of heavy atoms not in the maximum common substructure (MCS) with a reference compound is below the specified threshold for both the compound and the reference. Generally, the starting ligand should be selected as the FEP reference. Selecting more than 1–3 reference ligands is discouraged, as the MCS calculation is performed for every generated molecule against each reference.

FEP References text and load button

Load one or more reference compounds for the FEP+ amenability calculation. Select compounds from the Project Table and click Load, or click Browse to select a SMILES file. The total number of selected reference entries is displayed. A warning appears if any selected entries are not ligands.

FEP Max HAC perturbation text box

Set the maximum number of heavy atoms allowed to change in a perturbation. This is evaluated as the number of heavy atoms in the compound or reference (whichever is greater) minus the number of heavy atoms in the MCS. The default value is 10.

Similarity score option

Select this option to calculate the Tanimoto similarity of the newly generated R-groups with the R-groups of up to 5 compounds from the Project Table. The similarity calculation excludes the common core. Compounds with a maximum Tanimoto similarity below 0.1 are filtered out. The Tanimoto similarity score for each selected reference structure is included in the output properties.

Calculate QED score option

Select this option to calculate the Quantitative Estimate of Drug-likeness (QED) score for each generated compound. The QED score ranges from 0 to 1, where higher values indicate more favorable drug-like properties. The score is included in the output properties. For more information on QED scores, see Bickerton et al.

Epik pKa option

Select this option to filter molecules based on the pKa of the R-group vector explored by the AutoDesigner job. Molecules are filtered if the pKa of hydrogens on a formally charged nitrogen atom falls outside the specified range. This filter can be used to modulate the pKa of basic amine-containing molecules.

pKa Min text box

Set the minimum pKa value for the Epik pKa filter. The default value is 6.

pKa Max text box

Set the maximum pKa value for the Epik pKa filter. The default value is 9.

Phase hypothesis option

Select this option and click Browse to select a Phase hypothesis file (.phypo) generated in Maestro. The final docked poses are scored in place against the pharmacophore hypothesis, and compounds that do not match the required minimum number of phase sites are filtered out. The Phase pharmacophore hypothesis can be set up using the Develop Pharmacophore Model Panel.

Min phase site to match text box

Set the minimum number of pharmacophore sites that a docked pose must match to pass the Phase hypothesis filter. The default value is 2.

Design rationale option

Select this option to score each generated compound for its ability to increase or decrease a selection of ADMET endpoints compared to the starting ligand, based on combinations of relevant physicochemical properties. Up to two endpoints can be selected, and for each endpoint you can choose to either increase or decrease the property compared to the starting ligand. The available endpoints are:

  • Blood-brain barrier permeability

  • CYP inhibition

  • CYP time-dependent inhibition

  • hERG inhibition

  • Ligand complexity

  • Ligand rigidity

  • Liver microsome stability

  • Metabolic stability

  • OAT2 inhibition

  • Off-target liability potential

  • Permeability

  • Pgp efflux

  • Plasma protein binding

  • Solubility

  • Volume of distribution (Vss)

Job toolbar

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

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