Computational & In Silico Methods

de novo design

De novo design means having a computer invent brand-new molecules from scratch to fit a target, rather than picking from a catalog of existing compounds. Instead of shopping for a key that might open a lock, you carve a fresh key precisely shaped to the lock's interior.

Classical structure-based de novo methods grow molecules directly inside a target's binding pocket, placing atoms or fragments to make favorable contacts such as hydrogen bonds and hydrophobic fits, then connecting them into a complete, chemically sensible molecule. Ligand-based approaches instead build molecules that match a pharmacophore or resemble known actives. Modern generative methods learn from large datasets to propose novel structures with desired properties.

The enduring challenge is that it is easy to draw a molecule that scores well and hard to make one that is both synthesizable and truly active. Early de novo tools were notorious for proposing exotic structures no chemist could readily build. Good modern practice couples generation with synthetic-accessibility checks and experimental validation, treating proposed molecules as starting ideas to be filtered, made, and tested.

Synthetic accessibility is the perennial reality check: a beautifully docked de novo molecule is worthless if no feasible synthetic route exists to make it.

Also called
de novo molecular design全新药物设计全新藥物設計