Search Overview
Targeted searches when you know what you’re looking for
Beyond the metadata filters, the platform offers four targeted searches that rank iBGCs by how well they match something you supply. Use them when you have a concrete starting point — a known enzyme architecture, a protein sequence, a compound structure, or an interesting cluster you want more of.
| Search | You supply | Finds iBGCs that… |
|---|---|---|
| Domain | Protein domain accessions | …contain (or lack) those domains, or match a whole domain architecture |
| Sequence | A protein sequence | …encode a protein similar to yours (phmmer alignment) |
| Chemical | A SMILES structure | …are predicted to make chemically similar products |
| Find similar iBGCs | An existing iBGC (right-click) | …are most like that cluster overall |
How searches behave
- They combine with filters. A search runs within whatever metadata chips are active. Add a taxonomy or biome chip to restrict a sequence or chemical search to a clade or environment.
- They run when you press Run Query (except Find similar, which runs immediately from the right-click menu).
- Sequence and chemical searches run in the background. They are submitted as jobs; the discovery platform shows progress and fills in the results when the job completes. Domain and Find-similar searches return promptly.
- Results are ranked by match quality. The roster automatically sorts by the search score and the similarity column adapts to the search type (a domain similarity, a bitscore, etc.). The Variables map gains matching axes.
Which search to use
- You know the enzyme machinery (e.g. a PKS ketosynthase plus a specific tailoring domain) → Domain search, in architecture mode for an ordered run of domains.
- You have a specific protein (a characterised enzyme, a hit from your own data) → Sequence search.
- You have a compound and want clusters predicted to make similar chemistry → Chemical search.
- You found one good cluster and want its closest relatives across the catalogue → Find similar iBGCs.