Back

Design and validation of a clinical whole genome sequencing-based assay for patient screening in a large healthcare system

Wagner, J. T.; Welle, J. T.; Lucas Beckett, I. A.; Emery, K. R.; Cosgrove, B. A.; Olszewski, K.; Wagner, N.; Bower, T. C.; Yuan, L. C.; Shull, E. M.; Jade, K.; Clemens, J.; Magis, A. T.; Campbell, M. B.; Gordon, O. K.; Bifulco, C. B.; Piening, B. D.

2025-07-17 genetic and genomic medicine Community evaluation
10.1101/2025.07.16.25331598 medRxiv
Show abstract

Population genetic screening is rapidly emerging as a key methodology in the clinical laboratory for detecting actionable genomic conditions. While current clinical methods are largely focused on targeted gene panels, the increasing efficiency of next-generation sequencing (NGS) platforms permits the use of whole genome sequencing (WGS) for routine clinical applications. The key advantage of WGS is that the complete genome produced by a single sequencing event can form the basis for a patients genomic health care record for reanalysis throughout a patients lifetime. Here, we developed a scalable clinical WGS-based lab developed procedure (LDP) for heritable disease gene testing and pharmacogenomics (PGx). We performed extensive validation across a range of blood, saliva, and reference specimens. The clinical deliverable for the WGS LDP was 78 genes associated with actionable genomic conditions and 4 PGx genes. The validation cohort consisted of samples from 188 study participants that were orthogonally sequenced at commercial reference laboratories and additional reference materials. The deployed LDP was then used to sequence over 2000 patients as part of a broader clinical implementation study ("Geno4ME"). We demonstrate that the WGS LDP has excellent sensitivity, specificity, and accuracy, thus supporting WGS as a viable method for broad clinical screening.

Matching journals

The top 3 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.