CAD for synthetic-biology binding networks
The design workbench for binding-network engineers. Sketch reactions, verify the response geometry your circuit can reach, and commit to the design that survives parameter drift — all on one canvas, before a single primer is ordered.
Hardware CAD let chip designers stop fabricating prototypes to find out what their circuit would do. Biocircuits Explorer does the same for binding networks: the polyhedron of every regime your circuit can occupy is computed up front, so you commit to the design that survives parameter drift — not the one that happened to work in the last experiment.
Deploy to a browser, install the native macOS app, or stay on the original form UI for scripted work. Workspaces and atlas libraries are portable JSON across all three.
The full design canvas, served from any modern browser. Best for teams behind a shared backend, classroom deployments, and remote collaborators on a managed server.
SwiftUI shell with an on-device project drawer and an app-managed backend. Best for individual designers who want workspace files on disk and an atlas library that never leaves the machine.
The original form-based interface, kept around for reproducible scripting, headless reruns, and benchmarking against the canvas workflow.
Each primitive is a live node on the canvas. Wire them up, run, and the workspace updates in place — the same iteration loop a hardware designer expects from EDA.
01 · Response-surface geometry
Enumerate the exact vertices, edges, and faces of the Reaction Order Polyhedron. Rotate the geometry, inspect supports, and read dominant orders directly off the surface — the same way an EDA tool exposes timing margin instead of a binary yes/no.
02 · Input-output verification
Walk the SISO paths your input/output pair traces as parameters change. Compare asymptotic and singular families side by side. Spot the regimes a circuit will live in, before a single wet-lab cycle confirms it.
03 · Tolerance analysis
Run 1D and 2D parameter scans with live plotting. Pin a crosshair, drag across the grid, and the result node updates without leaving the canvas. The same tolerance loop a board designer runs against component variance — except the variance here is biology.
04 · Robustness under noise
Sample parameter space stochastically and overlay the cloud onto the polyhedron. Robust regimes show up as dense clusters; fragile ones are sparse outliers. Pick the design that lands in a robust regime under cell-to-cell variation — not just the one that looks clean on the schematic.
05 · Inverse design, AI Import, remote compute
Atlas-backed inverse design searches a library for networks that realize a target behavior. AI Import turns a paper or notebook into a populated analysis chain in one click. Cloud Compute offloads heavy atlas builds to AWS Batch. Workspaces stay versionable JSON across all of it.
Four steps that follow the same path in the browser, the native macOS shell, or the classic form UI. Workspaces and atlas libraries stay portable between them.
Type the reactions, Kd values, conserved totals, and observable species — or have AI Import extract them from a paper in one pass.
The model-builder turns the sketch into the matrices N and L, enumerates the regime vertices, and hands you a session every downstream node reuses.
Walk SISO paths, run parameter scans, sample the parameter cloud, and search the atlas. Exercise every regime your circuit can reach before you commit to a build.
Export the workspace as portable JSON, hand off SQLite atlas libraries, and reproduce the same analysis on a teammate's machine — or on a CI worker.
Open the web app for instant access, or install the macOS app for on-device projects and an app-managed backend. Both ship with the full canvas.
Web version
Zero install. Best for shared deployments, classroom sessions, and reviewing teammates' workspaces against a managed backend.
Open Web AppmacOS version
SwiftUI shell with an on-device project drawer, document files, and an app-managed Julia backend. Packaged builds ship through GitHub Releases.