Workflow builder for computational biochemistry

Designing for scientists

Most of the work was understanding the domain. What a ligand is, why docking runs in shards, and what makes a hit worth a second look. Scientists think in molecules and binding scores, not jobs and nodes, so results come first.

Start from a pipeline

Horus ships with 34 documented pipelines, like AutoDock Vina docking, Boltz-2 screening and GROMACS MD setup. Using one copies it into your workflows to change on the canvas. Any task can be saved as a reusable block, and a published workflow drops into another as a subworkflow.

Trying keys in a lock

Computational biochemists look for new medicines in simulation, long before the lab. They test thousands of small molecules against a disease protein to see which ones fit, like trying keys in a lock. The math runs for hours across dozens of machines, so Horus runs it on the cluster they already have, and if one piece fails, only that piece reruns.

Engineering lift

• Cloud machines spin up when a run reaches the step that needs them, and shut down when it ends, even on failure or cancel.
• Runs take hours and outlive the browser tab. Task and shard status streams from the backend, so the canvas, tasks table and inspector catch up after a reload.
• The canvas shares the open-source runtime's model. A node is a task, an edge is a typed artifact, and plugins add new targets and task kinds. Horus runtime on GitHub