BioBricks.ai: a versioned data registry for life sciences data assets

Area of Science:

• Biomedicine
• Public Health
• Bioinformatics
• Data Science

Background:

• Biomedical and public health researchers face significant delays due to manual data acquisition, cleaning, and integration from various sources.
• This manual process leads to inefficiencies, slow discovery, and inconsistent analytical pipelines.

Purpose of the Study:

• To introduce BioBricks.ai, an open, centralized repository designed to streamline access to and integration of public biological and chemical datasets.
• To provide modular, version-controlled data packages (bricks) with integrated extract-transform-load (ETL) pipelines.

Main Methods:

• Developed BioBricks.ai as a DVC Git repository system for public datasets.
• Implemented a package-manager-like interface for data installation, dependency management, and updates.
• Established a unified backend for data delivery.

Main Results:

• The current release offers over 90 curated datasets across genomics, proteomics, cheminformatics, and epidemiology.
• Combining bricks programmatically allows for rapid assembly of multi-dataset analytic cohorts, reducing time from days to minutes.
• Demonstrated significant acceleration in data access and integration compared to traditional methods.

Conclusions:

• BioBricks.ai accelerates data access, enhances workflow reproducibility, and lowers barriers to integrating heterogeneous public datasets.
• The platform fosters community contributions and reduces redundant engineering efforts by treating data as version-controlled software.
• Future work will expand data coverage and automated provenance tracking to improve FAIR data practices.