Skip to main content
The Problem You Are Trying to Solve
“I have a moderately active small molecule, and I want to improve it across multiple dimensions (potency, selectivity, developability, etc.) by exploring scaffold decoration and scaffold hopping, while preserving key motifs that drive favorable target interaction.”
Multi Parameter Optimize A Moderately Active Molecule Via Scaffold Decoration And Scaffold Hopping
At this stage, you’re no longer in discovery mode, determining what chemistries work for certain use cases, you’re deciding how to evolve the chemistry within certain chemical space constraints without breaking what makes the molecules active. This is traditionally difficult because:
  • Multi-parameter goals often conflict (e.g., potency vs solubility vs lipophilicity) and optimizing structure for one property usually will do the opposite for other properties you are concerned with
  • Scaffold hopping can accidentally remove the very interaction features responsible for binding
  • Minor changes can unexpectedly shift binding mode, strain, or developability risk
This workflow is designed to help you explore broader chemistry while staying anchored to the motifs that matter. Solution
This workflow enables users to run multi-parameter optimization (MPO) using Revilico’s generative chemistry suite, specifically leveraging Scaffold Decoration and Scaffold Transformations / Diversification to propose new analogs and scaffold hops that retain key interaction motifs. The primary engine chain is: QSAR and Pharmacophore Analysis → De Novo Library Generation (Scaffold Decoration / Transformations) → Docking → Molecular Optimization (using a variety ofMPO stages). This creates a controlled loop where:
  • You define what must be preserved (motifs + interaction features)
  • You generate scaffold-level hypotheses that respect those constraints
  • You triage structurally (docking) and iteratively improve multi-objectives (Molecular Optimization)
You optionally validate top candidates with MD / FEP when confidence needs to be higher What Data Do I Need to Provide?
Required
  • Starting molecule(s) as SMILES (moderately active lead(s))
  • Target protein structure (experimental or predicted)
  • A clear statement of what “better” means (your MPO priorities)
  • You must generally know structure activity relationships of your compounds so you know what scaffolds and motifs you must preserve when running optimizations. This can be done through virtual or experimental screening and co-crystal structure determinations
Recommended
  • Known binding pose(s) or docking results for the starting molecule
  • A description of the key motifs to preserve (functional group, ring system, H-bond pattern, charge center, etc.)
  • Any known constraints or liabilities (substructures to avoid, MW limits, LogP/TPSA ranges, etc.)
Optional
  • Experimental activity / selectivity data (improves scoring and QSAR usefulness)
  • A set of related analogs (helps define SAR and guide similarity constraints)
Workflow
  1. Define the Set to Preserve: Motifs + Interaction Features
Before exploring new scaffolds, establish what cannot be lost. On Revilico, users typically:
  • Run Docking (Static or Flexible) on the starting molecule to confirm a plausible binding mode
  • Use Pharmacophore Analysis to capture the binding-critical interaction features (donor/acceptor patterns, hydrophobics, charge centers, spatial arrangement)
  • Identify the motif(s) that should be preserved during decoration/hopping (e.g., hinge binder, charged anchor, aromatic stacking group)
  • If experimental data is already obtained at this stage, you can re-run the compounds in docking/pharmacophore engines to gauge target engagement across certain motifs and in QSAR modeling to extract necessary substructures that drive activity.
This step produces a clear “motif + interaction hypothesis” that will guide scaffold decoration/hopping and later scoring decisions. Integration note: If you have historical activity/ADMET data, you can optionally use QSAR Modeling here to help identify which structural features correlate with activity or liabilities.
  1. Generate Decorated Analogs Around the Core Scaffold
Now expand around your existing scaffold while preserving the core. Use De Novo Library Generation: Scaffold Decoration to:
  • Keep the scaffold fixed
  • Specify attachment points
  • Generate diverse R-group combinations that explore chemical space around your motif-preserving core
This step is ideal when:
  • You believe the core scaffold is correct
  • You want to systematically explore substituents to improve MPO dimensions (potency + solubility + stability + etc.)
This step results in a scaffold-preserving analog library that explores R-group space broadly and efficiently.
  1. Perform Scaffold Hopping / Core Replacement While Preserving Motifs
If your current scaffold is limiting (e.g., poor developability, IP issues, metabolic liabilities), move from decoration into scaffold hopping. Use De Novo Library Generation: Molecular Optimization modes (e.g., scaffold-based transformations or broad scaffold diversification) to propose scaffold hops that:
  • Retain key interaction motifs (pharmacophore features)
  • Explore alternate cores that could improve MPO properties (developability, selectivity, stability)
This step is ideal when:
  • You want new chemotypes (not just close analogs)
  • You suspect your current core is a liability but the binding motif is correct
This will create a scaffold-hop library: new cores with motif-preserving features and medicinally relevant chemistry. Integration note: If your organization has a proprietary chemical style or modality constraints, this is a great point to optionally use Custom Model Training to bias the generator toward “native” project chemistry.
  1. Re-score for Binding Plausibility
Now quickly remove structural false positives and prioritize binders, by running Docking on your generated library:
  • Start with Static Docking if the library is large
  • Use Flexible Docking for a smaller set or when binding-site residue movement matters
What you’re looking for
  • Plausible binding poses that preserve key interactions
  • Strong scores without obvious steric clashes
  • Consistency in top poses (not just one lucky pose)
This step will produce a downselected set of motif-consistent, pose-plausible candidates for deeper MPO refinement.
  1. Multi-Parameter Optimization Loop
With promising scaffold variants in hand, switch from “generate broadly” to “optimize deliberately.” Use Molecular Optimization Engine to run MPO by:
  • Defining multi-objective scoring (physicochemical bounds + predicted activities + penalties)
  • Using staged optimization to sequentially prioritize constraints (e.g., Stage 1: maintain motif + binding plausibility; Stage 2: improve developability; Stage 3: refine potency/selectivity)
This step is where you:
  • Balance tradeoffs (potency vs LogP vs PSA vs MW vs liabilities)
  • Keep the model anchored to reasonable chemistry (avoid drifting into unrealistic structures)
  • Control exploration vs conservatism using similarity settings, diversity penalties, and scoring structure
This engine will give you an MPO-optimized candidate list with transparent scoring breakdowns.
  1. Validate the Best Candidates with Dynamics and Energetics (Optional)
When you have a small shortlist and need higher confidence before synthesis:
  • Use Protein-Ligand MD to confirm binding stability and rule out pose artifacts
  • Use ABFE/RBFE Calculation to get higher-fidelity energetic ranking within close chemical series
This is especially valuable when docking can’t distinguish subtle improvements among similar candidates, and will provide you with a final shortlist supported by structure, stability, and (optionally) binding free energy confidence Results
  • A motif-preserving scaffold-decorated and scaffold-hopped candidate set
  • MPO-optimized compounds ranked across potency and developability priorities
  • A defensible shortlist of improved hypotheses ready for synthesis or experimental validation
  • Optional physics-based validation to reduce failure risk downstream
Now what? I have MPO-optimized scaffold variants, but what’s next?
  • At this point you can continue this cycle until you have a short list of compounds that you’d like to take into synthesis. The chemical hypotheses you have should be rigorously validated computationally.
  • For other parameters you’d like to optimize for, you can utilize any other Revilico Engine
Why Revilico?
Revilico makes scaffold-level MPO practical by connecting:
  • motif definition (Pharmacophore Analysis)
  • broad exploration (Scaffold Decoration + Scaffold Transformations)
  • fast structural triage (Docking)
  • deliberate multi-objective refinement (Molecular Optimization MPO stages)
  • optional high-confidence validation (Protein–Ligand MD, ABFE/RBFE)
This allows teams to explore chemical space aggressively without losing scientific control, and to produce candidates that remain interpretable, testable, and aligned with real drug discovery constraints.