Harnessing the power of PRS by combining standardized reporting with Allelica’s software
This week, scientists from the Clinical Genome Resource (ClinGen) Complex Disease Working Group and the Polygenic Score (PGS) Catalog published a paper in Nature introducing the Polygenic Risk Score Reporting Standards (PRS-RS). The work outlines the minimum set of data to allow scores to be faithfully interpreted and transparently evaluated.
An update of the now 10 year-old (and now rarely used for PRS) Genetic Risk Prediction Studies (GRIPS) statement, the PRS-RS lay out many of the essential pieces of information that are required to understand how and why a PRS was developed. These include information such as the background and motivations behind the score development, details of the demographics, ancestry and characteristics of the populations used to develop and validate the score, and the performance metrics required to provide objective assessments of the predictive performance of a PRS, its limitations and potential clinical implications.
As a company devoted to democratizing the use of PRS in healthcare, Allelica fully supports these recommendations and we welcome them as a way of ensuring that our field can provide the robust evidence base needed for the further adoption of PRS in healthcare.
In fact, thanks to a bit of advance warning — we’ve been working hard behind the scenes to make the outputs of our products align with these recommendations. Our Software as a Service is made up of three core components: the Discover module, which allows users to develop their own PRS; the Validate module, where users can test a PRS on a new dataset; and Predict, our main engine for computing PRS on individual level data at scale. Each of these modules works with the PRS-RS firmly in mind.
We have always believed that the best way to develop a PRS is to try multiple different algorithms and compare their output. So when deploying Discover and Validate, users are provided with multiple performance metrics, such as the Area Under the Receiver Operator Curve (AUROC), which is a measure of a score’s risk discrimination, and the Odds Ratio per Standard Deviation, a measure of its association with disease. Our end-to-end solution is built to ensure that users either provide as input or get as output all of the information required to follow the PRS-RS.
Armed with some data and Allelica’s software, users will find it easy to develop and validate their own scores and provide all the necessary information to adhere to these standards, which have been developed by the community, and to have all the required content to share their scores openly though the PGS Catalogue.
These standards help researchers to both develop a systematic framework for showing the validity of a PRS and provide an objective way to assess its predictive accuracy. As noted by the authors, this ‘will further facilitate the curation and expert annotation of published PRSs as we move towards widespread clinical use’. Importantly, the framework also provides the necessary information to provide the all important context to a score: the disease outcome the score aims to predict, the populations that the score is suited to, and perhaps most importantly, a way of articulating the intended purpose of a score, for example as a potential risk predictor for disease.
While additional information is needed to augment the PRS-RS for researchers aiming to document the clinical utility of PRS, these standards are an important milestone for the community and will provide the crucial structure to collect the evidence required to ensure that preventative medicine and healthcare can fully exploit the incredible potential of genomic data.
To demo our PRS software, contact us at info@allelica.com or book a free PRS consultation today: https://calendly.com/allelica/free-prs-consultation
