Comparative Analysis for the Performance of Long-Read-Based Structural Variation Detection Pipelines in Tandem Repeat Regions

Abstract

There has been growing recognition of the vital links between structural variations (SVs) and diverse diseases. Research suggests that, with much longer DNA fragments and abundant contextual information, long-read technologies have advantages in SV detection even in complex repetitive regions. So far, several pipelines for calling SVs from long-read sequencing data have been proposed and used in human genome research. However, the performance of these pipelines is still lack of deep exploration and adequate comparison. In this study, we comprehensively evaluated the performance of three commonly used long-read SV detection pipelines, namely PBSV, Sniffles and PBHoney, especially the performance on detecting the SVs in tandem repeat regions (TRRs). Evaluated by using a robust benchmark for germline SV detection as the gold standard, we thoroughly estimated the precision, recall and F1 score of insertions and deletions detected by the pipelines. Our results revealed that all these pipelines clearly exhibited better performance outside TRRs than that in TRRs. The F1 scores of Sniffles in and outside TRRs were 0.60 and 0.76, respectively. The performance of PBSV was similar to that of Sniffles, and was generally higher than that of PBHoney. In conclusion, our findings can be benefit for choosing the appropriate pipelines in real practice and are good complementary to the application of long-read sequencing technologies in the research of rare diseases.

Mapping structural variants to rare disease genes using long-read whole genome sequencing and trait-relevant polygenic scores

“…When comparing SV type across sexes, frequencies were similar and within 5% variability between males and females for length and count. The frequency and length of rSVs in our study closely correspond with population frequencies seen in other studies (Kosugi et al 2019; Guo et al 2021).…”

Is an SV caller compatible with sequencing data? An online recommendation tool to automatically recommend the optimal caller based on data features

“…In summary, different strategies investigate various variant signals (values and/or distributions) in sequencing data and can deal with diverse sequencing data with different signals and their distributions. Furthermore, some empirical studies (Luan al., 2020;Kosugi et al, 2019;Guo et al, 2021) have been conducted to validate this phenomenon. A set of popular callers is compared on some benchmarking datasets in these studies, and the results showed that most callers have an edge for specific data.…”

“…According to this study's testing, some features, such as read length and sequencing depth, are useful, while others are not. Furthermore, some review studies have proposed some sequencing data features, such as the size of SVs and the proportion of SVs in tandem repeat regions, which have been shown to differentiate the performance of SV callers (Kosugi et al, 2019;Zook et al, 2020;Guo et al, 2021).…”

“…In genomics and bioinformatics, calling structural variants (SVs) from sequencing data is a somewhat straightforward topic (Handsakeret al, 2011;Northcott et al, 2012;English et al, 2014;Cao et al, 2014;Guan and Sung, 2016;Chiang et al, 2017). Tens of review papers (Seo et al, 2016;Fang et al, 2019;Kosugi et al, 2019;Amarasinghe et al, 2020;Luan et al, 2020;Zhao et al, 2020;Zook et al, 2020;De Coster et al, 2021;Guo et al, 2021) highlight SVs as important biomarkers and routinely identify them in various fields. Therefore, many SV callers have been developed to detect SVs (Stancu et al, 2017;Gong et al, 2018;Sedlazeck et al, 2018;Wenger et al, 2019;Jiang et al, 2020).…”

Advancing genomic technologies and clinical awareness accelerates discovery of disease-associated tandem repeat sequences

“…Recent studies highlighted the need for LRS-specific algorithms for analysis of LRS data ( DeJesus-Hernandez et al 2021 ; Guo et al 2021 ; Miller et al 2021 ). While various combinations of base-caller algorithms and tandem-genotypes could provide estimates for the ABCA7 VNTR length, expanded alleles reaching more than 10 kb were better captured by a newly developed algorithm, NanoSatellite, which directly assesses electronic current data such as that obtained by nanopore sequencing ( De Roeck et al 2019 ).…”