Protocol Workflows | BioLM Docs

Orchestrating molecular design workflows

Protocols let you describe a multi-step molecular design job — chained model calls, gather/aggregation steps, filters, and structured outputs — as a single declarative YAML file. You author and validate the YAML locally, then either run it on your machine (Local Protocol Profile) or submit inputs to a registered slug on the BioLM platform.

This section is split into focused guides:

Choose local vs hosted

	Local execution	Hosted execution
Where it runs	Your machine (`biolm[pipeline]`)	BioLM platform servers
Entry points	`biolm protocol run-local`, `Protocol.execute()`	`biolm protocol run SLUG`, `run_protocol()`
Input format	`--input key=value` or Python `dict`	JSON file keyed to protocol `inputs`
Feature set	ApiTask DAG, `response_mapping`, `outputs[]` selection	Full protocol (gather, foreach, task-output expressions, …)
Best for	Fast iteration, offline dev, CI smoke tests	Long runs, sharing by slug, full platform features

See Local Protocol Profile v1 for the exact local feature matrix.

Guides in this section

Authoring and Validating Protocols — scaffold, inspect, and validate YAML
Running Protocols Locally — run-local, Protocol.execute(), outputs selection
Running Protocols on the Platform — submit, wait, download, MLflow logging
Local Protocol Profile v1 — Local Protocol Profile v1 reference

Choose local vs hosted

Guides in this section

See also