NGS Benchmarking v1.8


Recent updates...
  • v1.6 - This web tool now supports Genome Reference Consortium Human Build 38

About

This tool is provided for clinical labs to assess the performance of their NGS workflows for calling germline variants. It is based on the recommendations of the GA4GH benchmarking team and uses Illumina hap.py v0.3.9, powered by RTG vcfeval, to compare a sample VCF to the NIST Genome in a Bottle NA12878 truth set.

You can use this tool to assess NA12878 sequencing data you have generated by supplying a VCF file containing your variants and a BED file to restrict analysis to regions covered by your panel.

Alternatively if you just want to assess your bioinformatics pipeline, a pair of FASTQ files are provided. These were generated from Illumina NextSeq paired end whole exome sequencing on the NA12878 sample. You can run these through your pipeline and upload the resulting VCF (no BED file necessary).

You will receive an email containing the output from hap.py, which includes a summary file containing recall (sensitivity) and precision, as well as more detailed results. See here for more detail about interpreting these results.

Our Results

As a reference point we are providing our results for the provided FASTQ files, which we ran through our pipeline using BWA MEM and GATK Best Practices. The summary is shown below, and the full result set can be downloaded here.

Type Recall (Sensitivity) Precision
SNPs 0.99476 0.99694
INDELs 0.9177 0.90731

Contact

If you have any questions or comments, please post in Slack

Instructions

  1. (OPTIONAL) Download paired end FASTQ files (~12GB) here.
  2. Process above FASTQs or your own NA12878 data through your pipeline.
  3. Submit the gzipped VCF file produced by your pipeline, along with your email address and an optional BED file to restrict analysis regions, using the form below. Files should not contain patient identifiers.
  4. Await an email containing a link to your results.
    WARNING: Email might be marked as spam, please whitelist moka.alerts@gstt.nhs.uk!!

Submit

Results will be sent to this address

Max file size: 200MB

Max file size: 200MB

Supports builds GRCh37 & GRCh38

PLEASE NOTE: Upload may be slow if large VCF files are being submitted. Please be patient!