Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Virtual Laboratory for Managing Computational Experiments

Eleni Adamidi, Panayiotis Deligiannis, Nikos Foutris, Thanasis Vergoulis

TL;DR

SCHEMA lab tackles reproducibility challenges in computational science by capturing rich experiment metadata and integrating containerized task execution with high-level experiment management. The approach combines a front-end and a back-end API that orchestrate tasks and workflows on Kubernetes via TESK, with components for quotas, experiments, and file handling to preserve provenance. Key contributions include a modular architecture, open-source availability, and a roadmap to RO-Crate export and broader workflow-language support. The work has practical impact by enabling scalable, transparent, and shareable computational experiments across diverse infrastructures.

Abstract

Computational experiments have become essential for scientific discovery, allowing researchers to test hypotheses, analyze complex datasets, and validate findings. However, as computational experiments grow in scale and complexity, ensuring reproducibility and managing detailed metadata becomes increasingly challenging, especially when orchestrating complex sequence of computational tasks. To address these challenges we have developed a virtual laboratory called SCHEMA lab, focusing on capturing rich metadata such as experiment configurations and performance metrics, to support computational reproducibility. SCHEMA lab enables researchers to create experiments by grouping together multiple executions and manage them throughout their life cycle. In this demonstration paper, we present the SCHEMA lab architecture, core functionalities, and implementation, emphasizing its potential to significantly enhance reproducibility and efficiency in computational research.

A Virtual Laboratory for Managing Computational Experiments

TL;DR

SCHEMA lab tackles reproducibility challenges in computational science by capturing rich experiment metadata and integrating containerized task execution with high-level experiment management. The approach combines a front-end and a back-end API that orchestrate tasks and workflows on Kubernetes via TESK, with components for quotas, experiments, and file handling to preserve provenance. Key contributions include a modular architecture, open-source availability, and a roadmap to RO-Crate export and broader workflow-language support. The work has practical impact by enabling scalable, transparent, and shareable computational experiments across diverse infrastructures.

Abstract

Computational experiments have become essential for scientific discovery, allowing researchers to test hypotheses, analyze complex datasets, and validate findings. However, as computational experiments grow in scale and complexity, ensuring reproducibility and managing detailed metadata becomes increasingly challenging, especially when orchestrating complex sequence of computational tasks. To address these challenges we have developed a virtual laboratory called SCHEMA lab, focusing on capturing rich metadata such as experiment configurations and performance metrics, to support computational reproducibility. SCHEMA lab enables researchers to create experiments by grouping together multiple executions and manage them throughout their life cycle. In this demonstration paper, we present the SCHEMA lab architecture, core functionalities, and implementation, emphasizing its potential to significantly enhance reproducibility and efficiency in computational research.

Paper Structure

This paper contains 18 sections, 4 figures.

Table of Contents

  1. Introduction
  2. Background and Related work
  3. System Overview
  4. High-Level Architecture
  5. Back-End: SCHEMA api
  6. Execution manager
  7. Quotas manager
  8. Experiment manager
  9. Files adapter
  10. API
  11. Integration and Data Flow
  12. Front-End: SCHEMA lab
  13. Availability & Resources
  14. Demonstration Scenarios
  15. Scenario 1: Submitting and Monitoring Individual Tasks
  16. ...and 3 more sections

Figures (4)

  • Figure 1: High-Level Architecture of SCHEMA api. This diagram illustrates how SCHEMA api acts as a gateway between users and the containerized execution environment by performing authentication, scheduling containerized tasks via TESK, and providing monitoring endpoints.
  • Figure 2: Internal architecture of SCHEMA-api
  • Figure 3: Overview of an indicative workflow with one input, two outputs and three volumes
  • Figure 5: SCHEMA lab interface for creating an experiment by selecting executed tasks.