D4.2: Technical specifications, in machine-actionable metadata formats, for common workflow interfaces

This repository contains technical specifications for OO-LD (Object-Oriented Linked Data) schemas designed to standardize input/output interfaces for common materials science workflows. Built atop JSON Schema and JSON-LD, these schemas combine structural validation with semantic annotation to support:

• Interoperability through JSON-LD contexts and IRIs

• UI auto-generation from JSON Schema definitions

• Code-agnostic interfaces that abstract over simulation backends (e.g., VASP, QE, CP2K)

• Engine-agnostic compatibility with workflow systems (e.g., AiiDA, pyiron, FireWorks)

OO-LD enables both humans and machines to construct, validate, share, and interpret workflow definitions and results in a unified, extensible, and FAIR-compliant manner.

Authors

• Edan Bainglass (PSI)
• Giovanni Pizzi (PSI)

Goal

To provide specifications for standard schemas of common materials science workflows, including inputs, processes, and outputs. We will first deliver input/output schemas generalizing the inputs and outputs of the AiiDA common workflows (ACWF). Future work will aim at generalizing the schemas of the workflows themselves, providing a standard language for driving common materials science workflows. We will also in the future develop UIs for building and submitting common materials science workflows automatically constructed from the schemas.

Achievement

We developed engine-agnostic pydantic schemas for input/output components for common materials science workflows leveraging a code-agnostic structure geometry optimization (relaxation) workflow. The models are enriched with semantic annotation and export functionality to OO-LD documents, supporting JSON-LD algorithms for data interoperability. We provide an example notebook demonstrating the use of OO-LD documents in transferring workflow input based on engine-agnostic schemas down to the AiiDA workflow engine. AiiDA uses the transferred data and metadata to drive a structure relaxation workflow, yielding output in its own format. The notebook further demostrates conversion back to an engine-agnostic output schema.

External links

• Common workflow schemas repo
• OO-LD specifications
• AiiDA Common Workflows paper
• AiiDA Common Workflows documentation

Acknowledgements

The PREMISE project is supported by the Open Research Data Program of the ETH Board.