Format

Send to

Choose Destination
Nat Biotechnol. 2019 May;37(5):555-560. doi: 10.1038/s41587-019-0054-x. Epub 2019 Mar 11.

Best practices for benchmarking germline small-variant calls in human genomes.

Author information

1
Illumina Cambridge Ltd, Little Chesterford, UK.
2
Real Time Genomics, Hamilton, New Zealand.
3
Ontario Institute for Cancer Research, Toronto, Ontario, Canada.
4
Department of Physiology and Biophysics, Weill Cornell Medicine, New York, NY, USA.
5
The HRH Prince Alwaleed Bin Talal Bin Abdulaziz Alsaud Institute for Computational Biomedicine, Weill Cornell Medicine, New York, NY, USA.
6
The Feil Family Brain and Mind Research Institute, Weill Cornell Medicine, New York, NY, USA.
7
The WorldQuant Initiative for Quantitative Prediction, Weill Cornell Medicine, New York, NY, USA.
8
Department of Biomedical Data Science, Stanford University School of Medicine, Stanford, CA, USA.
9
Illumina Inc., San Diego, CA, USA.
10
Center for Devices and Radiological Health, FDA, Silver Spring, MD, USA.
11
Office of Health Informatics, Office of the Commissioner, FDA, Silver Spring, MD, USA.
12
Invitae, San Francisco, CA, USA.
13
DNAnexus, San Francisco, CA, USA.
14
Veritas Genetics, Danvers, MA, USA.
15
Broad Institute, Cambridge, MA, USA.
16
Bioinformatics Core, Harvard T.H. Chan School of Public Health, Boston, MA, USA.
17
Joint Initiative for Metrology in Biology, Stanford University, Stanford, CA, USA.
18
Material Measurement Laboratory, National Institute of Standards and Technology, Gaithersburg, MD, USA. jzook@nist.gov.

Abstract

Standardized benchmarking approaches are required to assess the accuracy of variants called from sequence data. Although variant-calling tools and the metrics used to assess their performance continue to improve, important challenges remain. Here, as part of the Global Alliance for Genomics and Health (GA4GH), we present a benchmarking framework for variant calling. We provide guidance on how to match variant calls with different representations, define standard performance metrics, and stratify performance by variant type and genome context. We describe limitations of high-confidence calls and regions that can be used as truth sets (for example, single-nucleotide variant concordance of two methods is 99.7% inside versus 76.5% outside high-confidence regions). Our web-based app enables comparison of variant calls against truth sets to obtain a standardized performance report. Our approach has been piloted in the PrecisionFDA variant-calling challenges to identify the best-in-class variant-calling methods within high-confidence regions. Finally, we recommend a set of best practices for using our tools and evaluating the results.

PMID:
30858580
PMCID:
PMC6699627
DOI:
10.1038/s41587-019-0054-x
[Indexed for MEDLINE]
Free PMC Article

Supplemental Content

Full text links

Icon for Nature Publishing Group Icon for PubMed Central
Loading ...
Support Center