vivarium-searchlisted

# vivarium-search — local sequence similarity search Find what a sequence is, or where it occurs, by searching it against a target with BLAST, DIAMOND, or HMMER — and hand back a clean, thresholded hit table plus a short read of what the hits mean. This is the "what is this gene / does this proteome carry a homolog of X / which domains are in this protein" workhorse. Everything runs locally in the `bio_tools` conda environment. Sequence search is light and fast, so this skill runs the search directly rather than handing back a command to run later — but it always records the exact command and tool version, because a hit table with no provenance is useless three months later when you write the methods. ## What the user should have at the end 1. A tidy TSV hit table with **named columns** (not raw outfmt-6 numbers nobody can read). 2. A short **interpretation**: how many hits pass threshold, the best hit(s), and any caveat (low coverage, borderline e-value, no hit). 3. One **provenance line**: tool + version + the exact command. ## Step 1 — read the task, pick the tool Infer (or ask) three things: are the QUERY and TARGET protein or nucleotide; how big is the target; is the goal a homolog search or a domain/profile search. The wrong tool either wastes time or misses real hits, so choose deliberately: | Goal | Query → Target | Tool | Why | |---|---|---|---| | Homolog search, protein, small/medium target | protein → protein | `blastp` | sensitive and exact; fine up to ~10⁵