Sean, in the Vogel Lab, pushes a stress-granule result to a read-only shared dashboard his collaborator Kate's lab reads with no Discourse Graph app — glance-and-trust for the PI, traverse-to-the-method for her students, and a Request back for more. Take the same machinery fully public and it's Brian, at the Quantum Biology Institute, publishing to the whole world. Two worked examples, one spine: push a result to a shared web interface — trustworthy and addressable at a URL, readable without any DG tooling.
A producer pushes a selected subgraph to a read-only shared web interface — public, or password-protected for a consortium — that a known collaborator's lab navigates in a browser, with a Request-back current. v0.4 rebuilds every screen on the lab's real prepared data — a Lattice-SIM study of stress-granule composition over time. Grounded in the Vogel × Kate sessions and the user-story canvas.
Sean picks the Evidence bundle, offers connected nodes, sets each node public / consortium / private (raw stacks gated, working notes withheld), and pushes a subset of his vault to the dashboard via KOI.
Open screen → 07A read-only web view at a URL — Graph · Kanban · Table, audience presets, node cards with provenance and a public/consortium badge — where Kate glances at the result and trusts it.
Open screen → 08A grad student follows the lineage to the segmentation choice (segment on the per-pixel SUM), opens the git / CSV pointers and the video walkthrough, hits the gated 80 GB raw stacks, and sends a Request for a follow-up.
Open screen → 09Kate runs one project umbrella across many labs — public vs. password-protected, preset views for funders and recruiting, and orphaned one-off results made discoverable instead of lost on a drive.
Open screen → 10 · evidence cardKate's Jun 10 ask, specced: one result-first card with the method folded in — collapsed for the PI glance, opened for the reproducer, a candidate awaiting promotion, and a comment thread that converts to a Request. Shown at all four states.
Open screen → 11 · meeting prepA "since last sync" diff Sean opens to prepare — what's on the agenda for the next meeting, what changed, and (folded) what hasn't. Self-set deadlines and an opt-in Follow for the PI. Mobile-first, for the trainee who arrives with only a phone.
Open screen →Take the same machinery fully public: no relationship, no shared tooling. Brian publishes a MagLOV2 result so anyone can read it cold at a URL, follow the pointers, request what's gated, and cite the exact version. One direction: producer → the world.
Brian selects the Evidence bundle and picks destinations (web DB · Jupyter Book · desci/nanopub) — setting each node public or request-access-gated, and keeping a working note private.
Open screen → 02Summary figure up front for the glancer; flip to full lineage — Study, Protocol, pointers — for the reproducer. Published, addressable, citable. One node, two depths.
Open screen → 03The result as a public web page at a URL — traversable lineage, pointer chips, a provenance byline to trust a stranger's figure, a frozen citation, and a path to request the gated raw data.
Open screen → 04Several independently-addressable Evidence bundles compile into one Jupyter Book micropublication; a model emerges above them; the narrative is AI-drafted, human-edited; the bundles stay live.
Open screen → 05The published graph as a public dashboard — every node a URL with provenance and a reuse count — plus citing a specific result with a frozen version pin, and seeing who's using your work.
Open screen → ↳ READThe full narratives behind the screens — the central Sean → Kate dashboard story and the QBI public story — and the normative spec other labs build from.
Open documents →