Benchmark for Detecting Low-Frequency Variants in the HG002 Genome
Author Information
Author(s): Camille A. Daniels, Adetola Abdulkadir, Megan H. Cleveland, Jennifer H. McDaniel, David Jáspez, Luis Alberto Rubio-Rodríguez, Adrián Muñoz-Barrera, José Miguel Lorenzo-Salazar, Carlos Flores, Byunggil Yoo, Sayed Mohammad Ebrahim Sahraeian, Yina Wang, Massimiliano Rossi, Arun Visvanath, Lisa Murray, Wei-Ting Chen, Severine Catreux, James Han, Rami Mehio, Gavin Parnaby, Andrew Carroll, Pi-Chuan Chang, Kishwar Shafin, Daniel Cook, Alexey Kolesnikov, Lucas Brambrink, Mohammed Faizal Eeman Mootor, Yash Patel, Takafumi N. Yamaguchi, Paul C. Boutros, Karolina Sienkiewicz, Jonathan Foox, Christopher E. Mason, Bryan R. Lajoie, Carlos A. Ruiz-Perez, Semyon Kruglyak, Justin M. Zook, Nathan D. Olson
Primary Institution: Medical Device Innovation Consortium (MDIC)
Hypothesis
Can we create a reliable benchmark for detecting low-frequency mosaic variants in the HG002 reference material?
Conclusion
The study successfully developed a benchmark for identifying low-frequency mosaic variants in the HG002 genome, demonstrating its utility for optimizing variant detection methods.
Supporting Evidence
- The benchmark includes 85 validated SNVs with VAFs between 5% and 30%.
- External validation confirmed the benchmark's reliability in identifying false positives and negatives.
- 13 out of 85 variants are located in medically relevant gene regions.
Takeaway
This study made a special list of tiny changes in DNA that are hard to find, helping scientists better understand and detect these changes in people's genes.
Methodology
The study used high-coverage sequencing data and a somatic variant caller to identify and validate mosaic variants in the HG002 reference material.
Potential Biases
Potential biases may arise from using specific sequencing technologies and reference materials that may not represent all genomic contexts.
Limitations
The benchmark may not include variants with allele fractions below 5%, which could limit its applicability for very low-frequency variants.
Statistical Information
P-Value
p<0.05
Confidence Interval
99% one-sided confidence interval ≥ 0.05
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website