Computational & In Silico Methods

pharmacophore model

A pharmacophore model is a three-dimensional sketch of what a molecule needs in order to be active, stripped of any particular chemistry. Think of it as a wanted poster that says, in effect, an active molecule must have a positive charge here, a ring that can stack there, and a hydrogen-bond donor at this distance, without saying which exact atoms provide them.

Each requirement is a pharmacophore feature, such as a hydrogen-bond donor or acceptor, a positive or negative charge, an aromatic ring, or a hydrophobic group, placed at a specific point in space with a tolerance. The set of these features, and the geometric distances between them, defines the model. It can be built from a single known protein-ligand complex or by overlaying several active molecules to find what they share.

Once defined, the model becomes a query for searching databases: any molecule whose flexible conformations can place the right features in the right places is a candidate, even if it has a completely different scaffold. This makes pharmacophore searching a fast route to scaffold-hopping. The caveat is that a sloppy or overly loose model retrieves many false hits, while an overly tight one misses real ones, so building and validating the model is the hard part.

Pharmacophore models can be structure-based, derived from a target's pocket, or ligand-based, derived only from known actives when no protein structure is available.

Also called
pharmacophore hypothesis药效团假设藥效團假設