BLAST Search

1. 01All five BLAST programs, against 549 databases, read live from the service registry
2. 02Every parameter the service advertises, rendered with its own description and defaults
3. 03Targets taken from the open document: features, the current selection, ORFs, or the whole sequence, validated before submission
4. 04Hit identifiers, organism, coverage, identity, E-value, bit score, frames and strand
5. 05The full pairwise alignment for each hit
6. 06A hit written back as a real feature at document coordinates, or imported as its own sequence
7. 07Exports of the result as TSV, FASTA, XML, the service report, or SVG
8. 08Job history that reopens a past run without resubmitting it

The client is the shared EMBL-EBI Job Dispatcher implementation, with parsing asserted against captured real jobs. Nothing about the database list is hardcoded: it is read from the service's own registry. The panel runs on @ports/comparison, @ports/jobs, @ports/networkLayout, @ports/phylo, @ports/queryAnchored, @ports/vecscreen.

Everything runs client-side. Files are parsed in the browser and projects are saved on this machine, so an unpublished sequence is never uploaded.

Does my sequence leave the machine?

Only when you press Search, and the panel says so before you do. The query is submitted to EMBL-EBI, which runs NCBI BLAST+. Everything else in the workbench runs locally.

How long does a search take?

It depends on the database: a protein search against Swiss-Prot returns in seconds, while a nucleotide search against the whole of ENA takes around ten minutes. The measured figures are shown next to the choice rather than left as a surprise.

Can a hit become an annotation?

Yes. A hit can be written back as a feature at the coordinates of the query region it aligned to, so the identification lands on the map instead of in a report.