Structural variations (SV) are critical genome changes affecting human diseases. Although many hybridization-based methods exist, evaluating SVs through next-generation sequencing (NGS) data is still necessary for broader research exploration. In this study, we comprehensively compared the performance of 16 SV callers and multiple NGS platforms (NovaSeq 6000, BGISEQ-500, GenoLab M, and MGISEQ-2000) using NA12878 whole genome sequencing (WGS) datasets.
The results indicated that several SV callers performed well relatively, such as Manta, GRIDSS, LUMPY, TARDIS, FermiKit, and Wham. Meanwhile, all NGS platforms have a similar performance using a single software. In size ranges of between 100 bp to 100 Kb, the six tools achieved higher performance. However, there was uneven performance in the shorter size and larger size regions, especially in size <100 bp. We evaluated the true positive (TP) deletions and found that NovaSeq 6000 platform detected the most TP deletions in all types, followed by MGISEQ-2000 and GenoLab M.
