Docking against TRPV1

Heat is not a property of a pepper. It is a property of a conversation between a molecule and a receptor: capsaicin binds TRPV1, the channel opens, and the nervous system reports something that feels like temperature. Everything the catalog does upstream — the capsaicinoid pathway, the enzymes that run it, the constructs that tune them — is in service of what happens at that binding site.

Which makes TRPV1 the natural target to dock against. When a pathway variation produces a capsaicinoid that is not quite capsaicin, the question that matters is whether the new molecule still fits the pocket, how well, and in what orientation. Docking is how that question gets asked cheaply, before anyone synthesizes anything.

A ligand and a target. Ligands come from the capsaicinoid set the platform already carries, from a catalog record, or from your own upload in standard small-molecule formats. The target defaults to TRPV1 — the receptor the whole catalog is organized around — and can be swapped for another structure, including one produced by the folding engine and sitting in the structure gallery.

Then pick the engine. AutoDock Vina is the default and the right answer most of the time: well understood, fast, and interpretable. DiffDock is available as an alternative when a deep-learning pose predictor is the comparison you want. Running both and looking at where they disagree is a legitimate use of the tool, and disagreement is informative rather than embarrassing.

Default engine

AutoDock Vina

Optional engine

DiffDock

Default target

TRPV1 receptor

Viewer

Mol*, in-browser

Returns

Ranked poses with scores

Engine class

GPU

Credit cost

5 credits

Classes

Standard, Performance, Max

Typical runtime

Minutes for a single ligand

Docking is GPU work, and GPU work runs on prepaid credits: $25 buys 250 credits, and one credit covers roughly one Standard job. A docking run costs 5 credits. Standard, Performance and Max classes are available when a run needs more machine behind it.

Treat the output as a hypothesis with a picture attached. A score reorders your candidates; it does not decide anything on its own. The reason the 3D view is given equal weight on this screen is that looking at where a ligand actually sits has saved more projects than reading a ranked list ever has.

Pharma and discovery groups use it as a triage step — many candidate molecules, one well-characterized receptor, and a need to decide which few are worth real work. Researchers studying the capsaicinoid pathway use it to ask whether an unusual pathway product still engages the receptor that makes it interesting. Flavour and hot-sauce R&D teams use it further upstream, where the question is about the character of a burn rather than its magnitude.

It sits naturally after structure prediction and variant scanning: fold the protein, scan for variants worth modelling, dock the molecules those variants would produce. Poses land back in the structure gallery alongside the model that produced them, which is where the next person on the project will go looking for them.