The Dockstore: enabling modular, community-focused sharing of Docker-based genomics tools and workflows

Area of Science:

• Genomics
• Computational Biology
• Bioinformatics

Background:

• Genomic datasets are rapidly expanding, making local analysis and traditional compute environments increasingly impractical.
• Large-scale genomics projects necessitate cloud-based infrastructure for data hosting and analysis.
• Challenges in data storage, distribution, and computational demands hinder local analysis of massive genomic datasets.

Purpose of the Study:

• To address the growing challenges in managing and analyzing large-scale genomic data.
• To develop a standardized platform for sharing and executing computational genomics tools.
• To improve the reproducibility and interoperability of genomics workflows.

Main Methods:

• Utilized cloud computing infrastructure for hosting and analyzing large genomic datasets, exemplified by the ICGC PanCancer Analysis of Whole Genomes (PCAWG) project.
• Employed Docker containers for packaging and sharing complex alignment and variant calling workflows.
• Developed Dockstore, a platform integrating Docker images with standardized, machine-readable descriptions for tool execution.

Main Results:

• Successfully processed over 5,800 whole human genomes using cloud environments and Dockerized workflows.
• Identified limitations in Docker for standardized tool description and execution.
• Launched Dockstore to provide a centralized registry for containerized genomics tools with standardized metadata.

Conclusions:

• Cloud-based infrastructure and containerization are essential for modern large-scale genomics.
• Dockstore enhances the sharing, reuse, and interoperability of genomics analytical tools.
• Standardization through platforms like Dockstore is crucial for advancing collaborative genomics research and promoting data sharing.