Skip to main content
Constraints work differently in Small-Molecule Design Projects compared to Sandbox. In Small-Molecule Design Projects, constraints are applied only to the predicted apo structure of your target protein - not to individual molecules in your virtual screening libraries.
Key Information: They only modify the target protein’s unbound structure.
If you need to test specific binding hypotheses (e.g., “does this warhead form a covalent bond?”), use Sandbox with constraints for those individual predictions. Then incorporate successful molecules into your Design Project.

The Design Projects Workflow

Understanding where constraints fit in the Small-Molecule Design Projects workflow is critical:
1

Create Target

Add your protein target sequence via FASTA, UniProt, RCSB PDB, or file import.
2

Apo Structure Prediction

Boltz predicts the unbound (apo) structure of your target protein automatically.
3

Verify Unbound Structure (Constraints Applied Here)

Review the predicted apo structure. If there are structural issues—incorrect domain orientation, problematic loops, or conformational problems - you can add constraints to guide the model toward a better structure.This is the only step where constraints are used in Design Projects.
4

Define Binding Pocket

Select residues that form your target binding site. Optionally provide probe molecules to help guide pocket prediction.
5

Virtual Screening

Screen your molecule libraries against the refined target structure. Individual molecules in these libraries cannot have constraints applied.

When to Use Constraints in Design Projects

Constraints in Small-Molecule Design Projects serve a quality control function - fixing structural predictions before virtual screening begins. Use them when:

Domain Misalignment

Multi-domain proteins predicted in non-native orientations that would block the intended binding site

Loop Conformations

Critical loops (e.g., DFG motif in kinases) predicted in inactive conformations when you need the active state

Cryptic Pocket Access

Allosteric or cryptic pockets that require specific residue positioning to remain accessible

Conformational States

Forcing a particular conformational state (open/closed, active/inactive) relevant to your screening hypothesis
“Only use this target if the structure is predicted well.”

How to Apply Constraints to Apo Structures

During the “Verify Unbound Structure” step, you’ll see the predicted apo structure with the option to add constraints.
Design Projects constraint workflow

Adding Constraints

Click ”+ Add Constraint” on the right panel to open the constraint editor. All three constraint types are available:
Constraint TypeUse Case in Apo Structure Refinement
BondForce disulfide bonds or covalent modifications not predicted correctly
ContactEnforce proximity between domains or secondary structure elements
PocketDefine where the binding pocket should form (alternative to manual pocket selection later)
Contact constraints are most commonly used in Design Projects to fix domain-domain interfaces or enforce known structural features from experimental data (crystal structures, cryo-EM).
The constraint setup process is identical to Sandbox - refer to the Setting Up Constraints guide for detailed instructions.

Best Practices for Design Projects

1

Start with the default prediction

Review the apo structure without constraints first. Only add them if you see clear structural problems.
2

Use experimental knowledge

Base constraints on crystal structures, homology models, or known domain interfaces.
3

Test in Sandbox first

If unsure whether constraints will help, run a quick Sandbox prediction with and without constraints to compare.
4

Validate pocket definition

After applying constraints and predicting the pocket structure, verify that the pocket is positioned correctly. Poor constraints can shift the pocket to the wrong location.

Sandbox vs. Design Projects: When to Use Each

ScenarioRecommended Workflow
Testing a specific covalent warhead hypothesisSandbox with bond constraints
Screening 50,000 molecules against a kinaseDesign Projects (no molecular constraints)
Fixing multi-domain protein orientation before screeningDesign Projects with contact constraints on apo structure
Validating a molecular glue ternary complexSandbox with contact constraints
Targeting a cryptic allosteric pocketDesign Projects with pocket constraints on apo structure
Exploring different binding modes of one moleculeSandbox without constraints (let model explore)

What’s Next?