Exciting news from our NGS-CN experts! The small variant calling benchmarking initiative by the NGS-CN and the GHGA is out and waiting for scientific feedback.
Publication Link: https://doi.org/10.12688/f1000research.140344.1
- Introduction of NCBench: A state-of-the-art, continuous benchmarking platform designed for the comprehensive evaluation of small genomic variant callsets. It looks into recall, precision, and false positive/negative error patterns.
- Revolutionary Implementation: NCBench operates as an ever-evolving open-source repository, showcasing the power of public free infrastructure. With the amalgamation of platforms like Github, Github Actions, and Zenodo, alongside recognized open-source tools, the possibilities are endless.
- Universal Application: One of NCBench’s strengths is its adaptability. It’s indifferent to the dataset used, capable of evaluating an endless number of callsets, and presents results visually and interactively.
- Real-World Testing: The team utilized NCBench to scrutinize over 40 callsets, derived from a variety of variant calling pipelines. These were executed on three exome datasets with varying enrichment kits and coverage levels. The results? While most pipelines displayed exceptional quality, NCBench highlighted nuanced systematic disparities amongst callers and datasets.
- Open Collaboration: NCBench is designed for community participation. Anyone can contribute a callset. A breakthrough for authors, it eradicates the need for repeated re-implementation of benchmarks for new publications. For readers, it offers a dynamic performance evaluation of tools and pipelines, ensuring up-to-date insights.
Delve into this pioneering work and stay at the forefront of genomic benchmarking!