Workflow Canvas — Guide

Tutorials

• Getting Started with Workflow Canvas
  • What is Workflow Canvas?
  • Prerequisites
  • Installation
  • Your First Pipeline
  • Next Steps
• Tutorial: Authoring a Method Script
  • Authoring a Method Script
  • Step 1 — Lay out the method directory
  • Step 2 — Write the script with the wfc-client decorator
  • Step 3 — Understand the contract underneath
  • Step 4 — Declare slots in method.yaml
  • Next steps
• Tutorial: Registering an Environment
  • Overview
  • Why environments exist
  • Choosing a backend
  • Referencing an env from a method
  • The ephemeral-container model
  • Requesting GPUs
  • Working inside an environment
  • Docker readiness and getting runnable
  • Next steps
• Tutorial: Writing Contracts
  • Writing contracts
  • Two tiers: module and method contracts
  • Declaring module contracts: file vs. CLI
  • Column validation: what’s inside the file
  • Worked example: from_params
  • Where contracts are enforced
  • Next steps

How-to Guides

• Registering Modules, Methods, and Samples
  • Registering a Module
  • Registering a Method
  • Registering a Sample
  • Registering from the Canvas
  • Next Steps
• Canvas Visual Builder
  • Overview
  • System Nodes
  • Builder Basics
  • Registry View
  • Pipeline Variables
  • History Basics
  • Next Steps
• Run & Inspect Results
  • Run & inspect results
  • Runs and outputs
  • Using the Canvas to inspect runs
  • Next steps
• How-to: Sweep Parameters & Fan Out
  • Sweep parameters and fan out
  • Parameter sweeps, variants & per-sample overrides
  • Fan-out and fan-in
  • Reusing values with pipeline variables
  • Next steps

Explanation

• Project Anatomy
  • Directory Structure
  • The Database
  • Configuration
  • Modules vs. Methods
  • The Run Workspace (.runs/)
  • Next Steps
• How a Run Executes
  • How a run executes
  • Snakemake is the orchestrator
  • The per-step unit: wfc run-step
  • The wfc ↔ method boundary: env vars + files
  • Where to go next
• Storage & Provenance
  • Storage & Provenance
  • The content-addressed cache is the only authoritative store
  • Archiving is deferred, and indexed by the database
  • What git tracks (and what it doesn’t)
  • Sharing results across machines
  • Where to go next
• Caching & Reproducibility
  • Caching & Reproducibility
  • Why a step re-runs (the cache key)
  • Lineage: tracing how a result was produced
  • Next steps