Antimicrobial resistance already contributed to an estimated 1.27 million deaths globally in 2019, and researchers project it could overtake cancer as a leading cause of death by 2050. The pharmaceutical response hasn't kept pace: only five new classes of antibiotics have reached the market since 2000 — a stretch researchers now call the "discovery void," driven by the high cost and low commercial return of traditional antibiotic development.
Antimicrobial peptides (AMPs) are one of the most active alternatives being pursued instead. They kill bacteria by physically disrupting cell membranes rather than targeting a single enzyme, which makes resistance emerge more slowly than it does against conventional antibiotics — and unlike most novel drug classes, AMPs are well-suited to data-driven discovery: large experimentally-validated databases already exist, and standard lab assays are cheap enough to validate AI-generated candidates quickly. The field's flagship reference, APD6, published its 2026 update in January with 5,188 catalogued peptides — 3,306 natural, 1,380 synthetic, 239 AI-predicted.
A list of sequences isn't a research tool
The bottleneck for most researchers isn't finding a peptide sequence — general databases already catalog thousands. It's everything that comes after: does this specific candidate have the physicochemical properties (molecular weight, net charge, hydrophobicity, the Boman index) that predict it'll actually work and not just sit in a spreadsheet? Is there a structurally similar peptide already characterized? What does the 3D fold look like before committing to a synthesis order?
PotatoPepDB is a curated database and analysis platform built specifically around antimicrobial peptides from Solanum tuberosum — potato — a plant source that's comparatively underexplored next to the frog-skin and insect peptides that dominate most general AMP databases. Beyond simple and advanced search by sequence, tissue, or activity, it runs the physicochemical calculations researchers actually need in real time (MW, pI, net charge, GRAVY, Boman index, instability index), offers BLAST-style similarity search, and renders 3D molecular structure directly from PDB/RCSB data — the analysis layer that turns a sequence list into something a lab can act on.
For a field this data-driven, the tooling around a database matters as much as the database itself. If your research still means exporting a sequence list and running the physicochemical math separately, that's exactly the gap a purpose-built platform like this closes.
