Skip to main content
How it works

From 2.39 million candidates
to a falsifiable in silico lead.

Bactaegion's pipeline turns a vast catalog of bacterial defense proteins into a few precise translational opportunities for broad-spectrum antivirals. Below is the seven-step path.

  1. 1.

    Anchor on the Bernheim 2026 Pasteur catalog

    GeneCLRDF — 2.39 million bacterial defense candidate proteins, 478,206 families, 12 systems experimentally validated. Source of truth, never duplicated.

  2. 2.

    Retain 5 V1 targets + 3 V2 ingested

    CBASS, Viperines, Pycsar, RADAR, Schlafen — chosen for their characterized host-directed analogs. Plus Ssp, DRT, Thoeris ingested via the biblio watch pipeline.

  3. 3.

    Fetch live from UniProt / AlphaFold / OpenAlex

    Sequences, structures, literature are fetched live by your browser. Bactaegion stores only pointers. Doctrine: "we create knowledge, we expose it, we squat what we can, we duplicate nothing".

  4. 4.

    Annotate via the 8 playable gestures

    Piano-roll annotation, BLAST alignment correction, pocket painting, structural superposition, operon scrolling, LLM ideation (BYOK), peer-review, curation — all 8 gestures from the engagement report are live.

  5. 5.

    Post a falsifiable hypothesis

    A lead = a mechanism summary + explicit falsifiability criteria + targeted proteins. Signed by your local DID, submitted via pull-request (or via Worker proxy if you don't have a GitHub account).

  6. 6.

    Peer-review with semantic taxonomy

    4 gauges + controlled vocabulary of 31 BACT-NNN labels + 200-char structured note. Median-aggregated, no leaderboard. Triplet verdict approve/reserve/caution.

  7. 7.

    Open the lead to the world (CC0)

    Retained leads enter the catalog under CC0. Anyone (academic lab, biotech, public health agency, generic compound maker) can pick them up to launch wet-bench validation. No patent ever.

Try the gestures Connect my engine Read posted leads