myExperiment: Transforming Scientific Workflows into Social Assets

TL;DR

Workflows are the "know-how" of modern science, yet they are often hidden in private scripts or lost in translation. This paper introduces myExperiment, a social networking paradigm for e-Scientists. By treating workflows as shareable scientific objects—much like photos on YouTube—the authors create a "gossip shop" and marketplace that accelerates discovery and fosters collaboration in high-throughput fields like Bioinformatics.

Positioning: This is a seminal "Community-Building" work that transitioned scientific computing from isolated execution environments to a collaborative Social Web.

Problem & Motivation: The "Plumbing" Trap

For years, the e-Science community viewed workflows merely as "plumbing"—the boring technical pipes connecting data to tools. The authors argue this view is fatal for two reasons:

1. The Cost of Invention: Workflows are "hard-won assets." Re-inventing a genomic annotation pipeline from scratch is a waste of human capital.
2. The "In the Wild" Reality: Bioinformatics services are volatile, scruffy, and lack formal contracts. Scientists need more than a compiler; they need a community-vetted map of what actually works.

The research's core insight is that e-Science is fundamentally "me-Science." Scientists are competitive; they will only share if it boosts their reputation or provides a clear competitive advantage.

Methodology: Social Networking for "Selfish" Scientists

The authors developed myExperiment not just as a repository, but as a virtual research environment (VRE).

Key Pillars of the Architecture:

• The Gossip Shop: Using familiar Web 2.0 metaphors (social tagging, ratings, and comments) to allow scientists to discuss the "invisible" tacit knowledge required to make a workflow run.
• The Marketplace: A "shopping" interface for workflows that avoids the dry, 1970s library-style catalogs.
• Workflow Mashing: A platform-agnostic gateway supporting Taverna, Kepler, and Triana, allowing for "meta-workflows" that bridge different ecosystems.

Experiments & Results: Beyond the Bench

The success of the Taverna/myExperiment ecosystem is evidenced by its scale:

• Integration: Access to over 3,500 operations, making it one of the largest functional collections in Life Sciences at the time.
• Community Adoption: Over 400 unique Taverna workflows were shared by the community before the formal platform even fully matured.
• Cross-Domain Expansion: The model proved robust enough to spawn pilots in Chemistry, Astronomy, and even Social Science.

The authors highlight that the "reward incentive" is the primary driver. By making workflows "citable" and "shareable," they transformed a private labor into a public scientific contribution.

Critical Analysis & Conclusion

Takeaway

The shift from "Technical Infrastructure" to "Social Infrastructure" is the defining contribution here. myExperiment proved that scientific tools must be "socially aware" to survive in the competitive academic landscape.

Limitations

• Interoperability: While the platform handles "metadata," the underlying workflow engines still struggle with execution-level compatibility (the "it works on my machine" problem).
• Incentive Alignment: While a "gossip shop" sounds ideal, the paper acknowledges the tension between open science and the "selfish scientist" who may fear being scooped.

Future Outlook

As we move into an era of AI-generated code and automated science, the principles of myExperiment—provenance, social validation, and reuse—are more relevant than ever. The next step for this field will likely involve "Self-Documenting" workflows powered by Large Language Models, but the social "trust" layer established here will remain the bedrock of scientific collaboration.