Open · Free · Community-driven
Every plastic-degrading enzyme,
in one atlas.
1.3 billion unique sequences. 64,730 enzyme families. 350+ characterized PETases. One interface to search, align, compare substrates, and trace evolution.
Start here
Four ways into the data
PETadex is a launchpad, not a paper. Pick the tool that matches your question.
Browse the family atlas
Explore 64,730 families in 2D UMAP space. Color by domain, phylum, or CATH component.
Find homologs by sequence
Paste a FASTA sequence; DIAMOND scans ~307M sequences in the Logan corpus and returns nearest neighbors with alignments and activity context.
Compare BHET activity
Side-by-side scatter of measured activity at 12.5 / 25 / 50 mM BHET. Click any gene to drill in.
Trace a family tree
Newick-rendered phylogenies for every family, with member tables and centroid sequences.
What's in PETadex
Six layers of data, one schema
| Layer | What's there | Tables / endpoints |
|---|---|---|
| Sequences | Curated amino-acid sequences from NCBI BLAST-NR + literature | fastaa · enzyme_fastaa · enzyme_taxonomy |
| Structures | Predicted & experimental structures via Molstar + 3DMol viewers | PDB · Molstar · 3DMol.js |
| Activity | Plate-reader BHET hydrolysis at three substrate concentrations | plate_data · plate_metadata · plate_activity_view |
| Provenance | Geographic origin, BioSample / SRA links, sampling dates | with_sra_and_biosample_loc_metadata |
| Embeddings | UMAP coordinates + CATH-domain colorings for every family | family_atlas (materialized view) |
| Phylogenies | Pre-computed Newick trees per family from S3-backed alignments | family/:id/tree |
Cite
Use it, share it, cite it.
PETadex is open-source under the MIT license. If a paper or pre-print uses our data, tools, or atlases, please cite the entry below.
@misc{petadex2026,
title = {PETadex: an open atlas of plastic-degrading enzymes},
author = {The PETadex contributors},
year = {2026},
url = {https://petadex.net},
note = {v0.4 (Apr 2026)}
}