Quality and clinical utility of genomic variants in complex diseases

Continuous improvements in high-throughput genomic sequencing over the past two decades have made it exponentially faster and cheaper, enabling its routine use in the clinic and scientific research. Genomic prognostic tools make use of personalised genomic data to aid clinical decision making and inform patients of disease outcomes, allowing enhanced tailoring of treatment beyond traditional prognostic tools, which are insufficient for understanding the nuances of individual complex disease cases. This relies upon accurate sequencing data and effective quality control. We have developed improved genomic prognostic tools for use in the clinic and demonstrate a novel method for quality control of genomic sequencing data with broad applicability.

Non-small-cell lung cancer (NSCLC) is the second most common cancer type in both males and females globally. Previous attempts to predict survival time for cancer patients have used genomic prognostic tools based on the burden of tumour mutations and neoantigens, but with limited success. We developed greatly improved classifiers of tumour mutation and neoantigenic burden showing strong 5-year survival differences between early-stage NSCLC patients. By using these together, we showed additional increases in prognostic efficacy, with the best survival group displaying a ~92% decreased risk of death in a 5-year period compared to the worst survival group.

To improve the accuracy of sequencing data for uses such as this, we developed the first tool for automatically cataloguing systematic sequencing biases for a sequencing pipeline, and we demonstrated its value in human and SARS-CoV-2 sequencing quality control with Illumina and Oxford Nanopore sequencing. We discovered and blacklisted a range of false positive variants, and investigated the causes of these. Identifying these errors contributed to multiple studies, altering research conclusions. We share these tools to provide continued improvements to genomic prognostics and sequencing accuracy affecting a wide range of fields.