Taming the reference genome jungle: the refget sequence collection standard

Area of Science:

• Genomics
• Bioinformatics
• Data Standards

Background:

• Reference genomes are crucial for genomic analysis but suffer from ambiguity and incompatibility.
• Inconsistent naming, undocumented differences, and lack of formal comparison mechanisms hinder genomic research.
• Existing standards do not adequately address the need for unambiguous representation and comparison of sequence collections.

Purpose of the Study:

• To introduce the GA4GH refget Sequence Collections (seqcol) standard for unambiguous representation, retrieval, and comparison of sequence collections.
• To enable precise identification and compatibility assessment of reference genomes and transcriptomes.
• To provide a scalable and reproducible solution to the reference genome compatibility crisis.

Main Methods:

• Developed a structured data schema for sequence collections.
• Implemented a canonical encoding algorithm for content-based, globally unique identifiers.
• Created a retrieval API and a comparison protocol for sequence collections.
• Applied the seqcol standard to 60 human and 36 mouse reference genomes using digest-based comparisons.

Main Results:

• The seqcol standard enables precise identification of sequence collections across decentralized systems.
• Analysis revealed substantial incompatibility and duplicate references among major genome providers.
• Identified consistent subsets of sequences and coordinate systems across different reference genomes.
• Demonstrated compatibility assessments beyond exact identity, including order-relaxed matches.

Conclusions:

• The refget seqcol standard offers a robust solution to reference genome incompatibility, enhancing transparency and reproducibility.
• Improved interoperability across genomic tools and datasets is achieved through the seqcol standard.
• The provided Python package, web API, and comparison interface support widespread adoption and application of the seqcol standard.