How to Use PocketXMol | Small Molecule Design

PocketXMol | Small Molecule Design is the structure-based drug-design entry point into PocketXMol. It focuses on three small-molecule workflows: de novo generation directly from a receptor pocket, fragment growing from a seed molecule, and fragment linking across user-defined fragment groups.

On Neurosnap, researchers upload an Input Receptor, choose Small Molecule Design Mode, define the pocket with either Pocket Reference Molecule or Pocket Center Override, and optionally provide an Input Seed Molecule plus Fragment Atom Groups for fragment-centered jobs. For seed-based workflows, PocketXMol preprocesses the seed molecule and strips explicit hydrogens before applying fragment atom indices. This setup is useful when a campaign wants to move from target structure to ligand ideas without conflating that task with docking or peptide-design inputs.

PocketXMol models molecular generation directly in three dimensions, conditioning the search on local atomic interactions inside a receptor pocket. In De Novo mode, the model samples new small molecules compatible with the selected site. In Fragment Growing and Fragment Linking, it preserves user-specified fragment context and generates new chemistry around or between those anchored groups.

On Neurosnap, Input Seed Molecule, Fragment Pose Mode, and Fragment Atom Groups determine how much of the starting fragment geometry is preserved for fragment-based workflows. Because PocketXMol strips explicit hydrogens from the seed molecule before fragment indexing, Fragment Atom Groups should be defined against the hydrogen-stripped atom numbering used at runtime rather than the original uploaded numbering. Small Molecule Atom Mean, Small Molecule Atom Std, and Small Molecule Atom Minimum shape the expected molecular size distribution during generation, while the pocket-definition fields still control where the model is allowed to place chemistry.

Researchers should interpret the outputs as pocket-conditioned ligand proposals. The value of the workflow is in narrowing chemical hypotheses to structures that are at least locally compatible with the receptor site before docking validation, medicinal-chemistry review, or synthesis planning.

• Design-only control surface: The service removes docking and peptide-specific inputs so the small-molecule workflow is easier to configure correctly.
• Supports both seed-free and seed-guided discovery: De novo generation, fragment growing, and fragment linking all operate within the same receptor-conditioned modeling framework.
• Pocket-aware chemistry generation: The receptor site is part of the generative setup rather than only a later scoring step.
• Medicinal-chemistry-ready outputs: Candidate structures can move directly into docking validation, analog review, and synthesis prioritization.