A researcher with a candidate peptide sequence wants one question answered fast: has something like this already been tested, and did it work? Scrolling through a spreadsheet of a few thousand entries doesn't answer that. A proper sequence search does, but only if the database has the search built in and not just the data sitting in rows.
A sequence list isn't a search tool
Plenty of antimicrobial peptide databases are really just a spreadsheet with a search box that matches on a name or a source organism. That's useful if you already know what you're looking for. It's useless if what you have is a novel sequence and you want to know which known peptides it resembles: near-identical matches are exactly the ones a name-based search will never surface.
Similarity search changes what the question even is
A BLAST-style alignment search compares a submitted sequence against the whole database and ranks results by actual similarity, not by any metadata a curator happened to attach. That turns "does anything like this exist" from a manual scan into a query that runs in seconds and comes back ranked. For someone screening a batch of candidate peptides against known activity, that's the difference between an afternoon and a week.
What actually helps
Sequence search built into the database itself, not a separate tool researchers have to export data into. Fewer format conversions, fewer versions of the dataset floating around, and a result that points straight back to the record with the activity and source data attached.
