Abstract

Traditional diagnostics tests measure a small numbers of analytes and have a short, largely automated laboratory workflow with a simple readout which can be used to create the report based on reference ranges associated with the measure analtyes. NGS assays are the polar opposite. The laboratory workflow can take multiple days in which samples are combined to create multiplexed samples which are sequenced together in a single sequencing run on the instrument. The data is then processed through complex multi-step bioinformatics pipelines. The pipelines may detect tens of thousands of variants in a single sample from which the one or two variants associated with the clinical phenotype need to be identified, possibly confirmed, and supported by literature evidence. As the cost of sequencing reduces, many diagnostic laboratories and beginning to invest in NGS diagnostics. Laboratories often use bioinformatics pipelines which have been developed for research use and are not validated for clinical use. In this talk, we will discuss strategies for evaluating and optimising these pipelines. We will also discuss how the sequencing data can be used to identify common problems such as accessioning errors, cross-contamination, instrument performance degradation, NGS protocol errors etc. Finally, while the majority of detected variants will have good quality indicators a small fraction will have borderline qualities. We will suggest efficient strategies for second method evaluation that can be used to confirm these variants.