Features

An overview of what Salpa offers for computational science workflows.

Visual Workflow Editor

The core of Salpa is a drag-and-drop node editor powered by ReactFlow:

• Canvas — Infinite workspace where you place and arrange nodes
• Node Library — Browse available nodes by category, search by name or hashtag
• Connections — Draw edges between node ports to define data flow
• Execution — Run individual nodes or entire workflows with one click
• Real-time updates — Live progress bars, streaming logs, and status indicators via WebSocket

Node System

Every computation in Salpa is a node — a self-contained unit with defined inputs, outputs, and execution logic:

• Two-level architecture — core.py contains pure science code; node.py wraps it for Salpa
• Isolated environments — Each node can have its own Python environment via Pixi (no dependency conflicts)
• Auto-detected execution — Nodes with pixi.toml run in an isolated subprocess; others run in-process
• Metadata-driven — meta.toml defines the node’s name, version, category, hashtags, and description

See Node System for details.

Node Marketplace

Discover and install community-built nodes without leaving the app:

• Browse nodes by category (molecular dynamics, structure prediction, analysis, utilities)
• Search by name or hashtag
• One-click install — Dependencies are handled automatically via Pixi
• Uninstall cleanly when no longer needed

Cloud GPU Computing

Run GPU-heavy workloads on cloud hardware without managing infrastructure:

• A100 and H100 GPUs — ESMFold on A100, Boltz-2 on H100
• Pay-per-use credits — No AWS/GCP account needed
• Transparent execution — Cloud nodes look and behave like local nodes
• Firebase authentication — Sign in with Google to access cloud computing

See Cloud Computing for details.

Multi-Platform Desktop App

Salpa runs natively on all major platforms:

The desktop app bundles everything needed — Python, Redis, and all backend services start automatically.

Execution Environments

Salpa supports multiple execution environments for different use cases:

• Pixi (Default) — Isolated Python environments per node. Zero dependency conflicts, automatic setup. Every node with a pixi.toml gets its own reproducible environment.
• Container (Coming Soon) — Docker/Podman support for maximum portability and reproducibility. Package entire computational environments into shareable container images.
• HPC / SLURM — Built-in SSH profile manager for remote cluster execution. Vendor-dependent configuration. Contact us for institutional integration support.

Workflow Management

• Save and load workflows as JSON files
• Export workflows for sharing or version control
• Duplicate workflows to create variations
• Templates — Start from pre-built workflow templates for common pipelines

FAIR & Reproducible Workflows

Salpa treats every workflow as a self-describing, executable record:

• Workflows as executable “Methods” sections — Instead of writing opaque prose, share a JSON file that captures every parameter, connection, and environment
• JSON export — A complete, human-readable record of the entire pipeline. Version-control it alongside your data
• Template sharing — Save workflows as templates for collaborators to reproduce your exact pipeline
• Environment pinning via Pixi — Each node locks its exact dependency versions, ensuring identical execution across machines and time
• Open source — Inspect the code, contribute nodes, and build on a transparent, community-driven platform

Settings and Customization

• Light and dark themes with glassmorphism UI
• Canvas backgrounds — Choose from 6 styles (dots, grid, gradient, mesh, and more)
• Configurable working directory per workflow
• Home page tips — Contextual tips powered by a remote channel feed

Job Monitoring

Track all running and completed jobs:

• Jobs page — View all jobs with status, progress, and timestamps
• Log Center — Centralized logs from all node executions
• Environment Manager — View and manage Pixi environments for installed nodes