Dan Turner SVP of the Applications Team at Oxford Nanopore Technologies, presented the update from the Apps Team. Turner summarized the use of Nanopore data for clinical whole-genome sequencing of rare diseases and cancer with the goal of identifying relevant variants. Turner explained that although rare diseases are rare by definition, there are over 3,000 different rare diseases based on work by Hudson Alpha. Dark regions are those that seem inaccessible to short read sequencing: typically repeat regions that are challenging to resolve. To benchmark their updates, they sequenced well known samples and provided the F1 recall/precision values. SNPs, indels, and SV had very high F1 values with Oxford Nanopore Technologies sequencing. Long reads made a difference with SNP and indel F1 values above 99. For SV, Turner noted that the F1 of 96% was due to the rather incomplete nature of the truth dataset. SV detection can be improved with kit 114 chemistry and even further with ultra-long read library prep and Pore-C proximity sequencing. Turner emphasized that genetic diseases are not simple. Genome-wide association studies (GWAS) can help test the effects of variants and SNPs to diseases. Genotyping large cohorts can help evaluate and determine polygenic risk scores from the data. Turner noted that ideally telomere-to-telomere sequencing would be ideal, but cost prohibitive and highly involved. SNP-chips are a snapshot of a genome and don’t find anything unknown. SNP-chips are also not detecting structural variants and methylation. Turner and colleagues think there is a good argument for the use of low-coverage (1x) of a genome. For gaps, imputation can be used to predict the variants that were not detected. Interestingly, shorter reads provided useful information. Skim-seq was developed for low-coverage whole-genome sequencing using adaptive sampling. To test this approach, they generated polygenic risk scores for a variety of conditions. Turner explained that using low coverage, more samples could be sequenced on a PromethION flow cell. Further, adaptive sampling could help sequence only regions of interest, reducing the need for multiplexing. The Applications Team continue to work on this approach, and Turner said the results presented were very recent.
