Interlaboratory evaluation of high molecular weight DNA extraction methods for long-read sequencing and structural variant analysis.

Background: Long-read sequencing technologies enable resolution of structural variants (SV) and long-range genome assembly, but require high molecular weight (HMW) DNA of both high quantity and quality to produce optimal sequencing results. New DNA extraction methods have been developed but these have not been assessed for use in routine testing. The interlaboratory study described here tested four commonly used methods: Fire Monkey, Nanobind, Puregene and Genomic-tip with a reference cell line containing known chromosomal alterations. Samples were assessed with commonly applied approaches for evaluating DNA purity and integrity as well as a method based on linkage using digital PCR. Sequencing performance was evaluated and the impact of extraction method on structural variant calling investigated.

Results: All methods generally produced samples of acceptable purity although yield varied considerably between laboratories. Library preparation and sequencing were successful for all four methods, with Fire Monkey extracts achieving the highest N50 values, Genomic Tip giving the highest sequencing yields and Nanobind, the highest proportion of ultra-long reads (> 100 kb). The dPCR assay with duplexes at 100 kb and 150 kb distances was predictive of ultra-long reads and provides a more quantitative read-out (% linkage) than pulse-field gel electrophoresis (PFGE) which varied in performance between instruments and gel dyes. Neither PFGE nor dPCR were predictive of the proportion of short reads (< 10 kb). Coverage was a key factor in the success of SV calling, but this was dependent on SV caller. Megabase scale SVs were challenging to analyse with SV callers and required confirmation based on coverage plots and mapping of junction sequences, and the findings of earlier studies were only partially confirmed.

Conclusions: This study highlights some of the challenges of HMW DNA extraction as well as the need for robust sample QC metrics to ensure optimal sequencing yield and read length which in turn influence the success of SV analysis. dPCR approaches for DNA integrity showed potential but require further development. As long-read methods are increasingly applied in routine settings such as clinical testing laboratories, cellular reference samples with well-characterised SVs are recommended as controls for the full long-read sequencing workflow.