GEMSTAR stands for Genetic Medicine Standards Tools And Resources. GEMSTAR is the suite of flexible, user-friendly outputs that TGMI has developed to help deliver safe, consistent, transparent genetic testing.
Genetic testing quality is essential
High-quality genetic testing is the cornerstone of safe genetic medicine. But to ensure genetic tests are of the right quality we need to decide the quality standards testing should meet. Then we need to develop tools that ensure testing meets the standards. And we also need benchmarking resources that show if, and how well, the standards are met.
The Quality Sequencing Minimum (QSM) standardThe quality of the DNA sequencing determines the quality of a genetic test. It has proved hard to communicate the quality of DNA sequencing simply and transparently. This has made it hard to set objective quality standards for DNA sequencing. In turn, this has it made it hard to set quality standards for genetic testing. To help solve this TGMI developed the Quality Sequencing Minimum (QSM). The QSM is a simple, comprehensive and transparent shorthand that labs can use to describe the sequence quality of their genetic test. Broad adoption of the QSM will help development of guidance and guidelines about the required sequence quality in genetic testing. You can find out more in our QSM publication and blog.
Genetic testing tools
We have developed five tools for use in genetic testing. All are freely available to download. We have published papers on each of the tools which describe their capabilities and uses in detail.
- OpEx is a press-and-play open-source exome analysis pipeline. We recently published a comparison of OpEx against GATK and Google’s new variant caller – DeepVariant. You can read the results here. You may also like to read our blog about exome sequencing and OpEx. We have included two of our other tools, CAVA and CoverView in OpEx.
- CAVA is a fast, lightweight tool to aid genetic variant annotations, interpretations and comparisons.
- CoverView is a fast, flexible quality evaluation tool for that helps users to achieve quality standards. Additionally, CoverView highlights sub-optimal data needing further attention.
- DECoN is a tool that detects a tricky type of variant called exon CNVs, that are too large to be robustly detected by general variant callers.
- ICR142 Benchmarker is our most recent tool. It uses the ICR142 NGS validation series to provide information on the sensitivity, specificity and false detection rate of variant callers. It also generates a one-page report that highlights key performance metrics and shows how performance compares to widely-used tools.
Genetic testing benchmarking
It is difficult to objectively check how well your genetic testing is performing. Most laboratories use internal data, which is often limited in size and scope. Or they use artificially generated data, which is useful, but cannot fully represent or replace human data.
We have assembled three benchmarking datasets that are available to any clinical, research or commercial user. They can be used to evaluate and improve genetic testing, and to show how well a test performs compared to the required standard, or tests from other providers.
- ICR142 NGS validation series. This dataset enables users to check how good their variant detection pipeline is at finding real variants (true positives) and at not finding variants that aren’t really there (false positives). The dataset has strong representation of insertions and deletions, which are a major cause of disease, but can be difficult to detect. ICR142 Benchmarker makes it easy to use the ICR142 series to evaluate and compare variant detection performance.
- ICR639 CPG NGS validation series. This dataset enables users to check how good their cancer predisposition gene testing pipeline is at detecting disease-causing variants. It includes representation of all the different types of variants, including those hardest to detect.
- ICR96 exon CNV validation series. This dataset enables users to evaluate and improve exon CNV detection. Exon CNVs are difficult to detect and many laboratories currently either do not detect them at all, or use a separate method, which adds time and cost.
- ICR1000 UK exome series. We have also made available exome data from 1000 people from the UK general population that are part of the unique 1958 Birth Cohort. Access to this dataset is through the 1958 BC committee.
We need more standards, tools and resourcesGEMSTAR has already proved useful to many. We believe the GEMSTAR components have high potential to become broadly used, which will help to improve and standardise genetic testing. However, GEMSTAR is only a start – we need many more standards, tools and resources. In particular, we urgently need more benchmarking datasets to objectively evaluate and compare genetic testing performance.
Most importantly, we hope GEMSTAR will help to drive transparency in genetic testing. We must all be transparent about the general and individual limitations of genetic testing. This openness will focus us on what needs to be improved in genetic testing and will help us to deliver safe, effective genetic medicine.