Vcf2db enables linked querying across multiple molecular data types
Bridging the Omics Divide: A Modular Relational Approach to Multi-Layer Biological Data Management
Databases
Summary
Large biological datasets from different types of molecular experiments are often stored separately, making it hard to search across them together. The authors created vcf2db, a system that organizes data by samples and connects different molecular layers, like DNA variations and gene activity, in a database. This approach lets users run combined searches using a common language, making analysis easier and more reproducible. They showed that vcf2db performs well for many DNA-focused queries and can easily add other data types without changing the original setup.
What this means in practice
- •For bioinformatics teams: Integrate and query genomic and transcriptomic data jointly via SQL to enable more comprehensive biological analyses.
- •For clinical data engineers: Streamline management of large-scale patient molecular data with modular schemas linking multiple omics layers by sample identifiers.
Authors
Alessandro Balestrucci, Donald Friggieri, Andrea Gariboldi, Panagiotis Alexiou
Abstract
Background: Rapid growth of high-throughput molecular data demands systems for efficient retrieval, integration, and scalability across omics layers. Traditional file-based workflows hinder cross-modal analysis and reproducibility because of fragmented storage and ad-hoc querying. Few existing tools for genomic variation data prioritize modular multi-omics integration. We developed vcf2db, a sample-centric relational framework modeling each omics modality as a distinct but linkable component centered on biological samples. This study evaluates whether this design delivers competitive genomic retrieval while enabling extension to additional molecular layers. Results: We implemented a proof-of-concept genomic schema and ingestion pipeline for annotated VCF data using the European subset of the 1000 Genomes Project (502 samples, 25 million variants). We benchmarked it against three established VCF-oriented tools on seven retrieval tasks: coordinate filtering, annotation-driven queries, genotype extraction, and aggregation. Under controlled conditions, vcf2db performed strongly on selective queries, often outperforming other systems for coordinate and annotation filters, and remained usable for genotype retrieval. Aggregation-heavy tasks were less efficient, indicating optimization targets. We also validated modular extensibility by adding a synthetic transcriptomic layer without modifying genomic tables, linking layers via shared sample identifiers. Conclusion: vcf2db supports cross-layer retrieval directly as SQL queries anchored on shared sample identifiers, enabling integrated multi-omics access that is difficult with file-based approaches.