ORSO

ORSO connects researchers to relevant high-throughput sequencing datasets by recommending related genomics data using primary read coverage, annotated metadata, and user interaction data to reveal biologically meaningful relationships.

Key Features:

• Data-driven social network: Uses user interactions such as favoriting datasets and following researchers to inform recommendations.
• Extensive dataset repository: Hosts over 30,000 datasets from major biomedical consortia.
• Recommendation system: Leverages user interactions alongside primary read coverage information and annotated metadata to compute dataset similarities and recommend related datasets.
• Graph-based exploration: Represents dataset associations as a graph whose topology groups samples from related biological systems.
• Validation and accuracy: Validated with an RNA-seq time course tracking differentiation of embryonic stem cells into cardiomyocytes, correctly labeling early time points as embryonic stem cells and late points as heart and muscle samples.

Scientific Applications:

• Dataset discovery: Facilitates locating relevant genomics data across thousands of high-throughput sequencing datasets.
• Cross-study comparison: Groups related samples to enable comparative analysis across datasets and biological systems.
• Support for basic and disease research: Connects researchers with appropriate sequencing datasets for basic biological studies and complex disease investigations.

Methodology:

Computes dataset similarity from primary read coverage information and annotated metadata, integrates user interaction signals (favoriting, following) into a recommendation algorithm, and represents associations via a graph-based network topology; validated using an RNA-seq time course of embryonic stem cell differentiation into cardiomyocytes.