SVIM: structural variant identification using mapped long reads.
Heller D, Vingron M. · 2019
Read SVIM: structural variant identification using mapped long reads. by Heller D, Vingron M. (2019) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/europmc-pmc6735718.txt.
Introduction
Heller and Vingron present SVIM, a tool for finding structural variants in genomes. Structural variants are changes larger than 50 base pairs. They affect more DNA than small variants and link to many diseases. Their size and ties to repeated DNA make them hard to spot with short reads. Long-read sequencing helps, but older tools left gaps.
SVIM collects, clusters, and combines variant signatures from long-read alignments. It separates five variant classes, including tandem and scattered duplications and inserted new elements. Tests show it detects variants with high sensitivity and precise positions. The tool aims to help researchers link these large changes to human diversity and disease.
Catalog record
| CREATOR | Heller D, Vingron M. |
|---|---|
| YEAR | 2019 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 50 KB |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
SVIM: structural variant identification using mapped long reads By David Heller, Martin Vingron, Inanc Birol Published in: Bioinformatics DOI: 10.1093/bioinformatics/btz041 ABSTRACT Abstract Motivation Structural variants are defined as genomic variants larger than 50 bp. They have been shown to affect more bases in any given genome than single-nucleotide polymorphisms or small insertions and deletions. Additionally, they have great impact on human phenotype and diversity and have been linked to numerous diseases. Due to their size and association with repeats, they are difficult to detect by shotgun sequencing, especially when based on short reads. Long read, single-molecule sequencing technologies like those offered by Pacific Biosciences or Oxford Nanopore Technologies produce reads with a length of several thousand base pairs. Despite the higher error rate and sequencing cost, long-read sequencing offers many advantages for the detection of structural variants. Yet, available software tools still do not fully exploit the possibilities. Results We present SVIM, a tool for the sensitive detection and precise characterization of structural variants from long-read data. SVIM consists of three components for the collection, clustering and combination of structural variant signatures from read alignments. It discriminates five different variant classes including similar types, such as tandem and interspersed duplications and novel element insertions. SVIM is unique in its capability of extracting both the genomic origin and destination of duplications. It compares favorably with existing tools in evaluations on simulated data and real datasets from Pacific Biosciences and Nanopore sequencing machines. Availability and implementation The source code and executables of SVIM are available on Github: github.com/eldariont/svim. SVIM has been implemented in Python 3 and published on bioconda and the Python Package Index. Supplementary information Supplementary data are available at Bioinformatics online. 1 INTRODUCTION A typical human genome differs from the reference genome at ∼4–5 million sites amounting to ∼20 million altered bases (1000 Genomes Project Consortium, 2015). These variations can be categorized into single-nucleotide polymorphisms (SNPs), small insertions and deletions (Indels) and structural variation (SV) affecting a larger number of base pairs. Typically, differences larger than 50 bp are considered SVs although definitions vary and sometimes overlap with those of Indels. Studies have shown that in human more base pairs are altered due to SV than due to SNPs (Redon et al., 2006; Weischenfeldt et al., 2013). Additionally, SVs are enriched 50-fold for expression quantitative trait loci when compared to SNPs (Sudmant et al., 2015). Unsurprisingly, SVs have a major influence on human diversity and are implicated in a wide range of diseases from autism and other neurological diseases to cancer and obesity (Sebat et al., 2007; Weischenfeldt et al., 2013). Consequently, the characterization of SVs is of major importance to human medicine and genetics alike. It can contribute to the early detection of disorders and can help to elucidate their underlying genetic and molecular processes (Gonzalez-Garay, 2014). In other organisms such as plants, SVs play an equally important role by driving phenotypic variation and adaptation to different environments (Saxena et al., 2014). Next generation sequencing has enabled the identification of SNPs and small Indels to a high resolution. SVs, however, are much harder to detect. One reason is that SVs encompass a diverse range of modifications. While SNPs are simple base pair substitutions, the term ‘SV’ summarizes many different phenomena. Typically, different classes of SVs are distinguished, such as deletions, inversions and insertions. Definitions for some of these classes vary in the literature. For the purpose of this work, we define six different SV classes which are visualized in Figure 1: deletions, cut&paste insertions, tandem and interspersed duplications, inversions and novel element insertions. The main drivers behind interspersed duplications in human are mobile element insertions, such as Alu, LINE1 and SVA elements. They duplicate using retrotransposition and in total represent ∼25% of all human SV (Stewart et al., 2011; Sudmant et al., 2015). DNA transposons, although now inactive in mammals (excepts bats) are active in plants and lower-order animals (Huang et al., 2012). They use a cut&paste mechanism to move in the genome and therefore motivated the inclusion of cut&paste insertions as a separate SV class. [Figure: Schematic overview of different SV classes. SVs can be categorized into deletions, cut&paste insertions, tandem and interspersed duplications, inversions and novel element insertions. Each SV class is depicted in an individual genome (lower line) when compared to the reference genome (upper line). The region being rearranged is marked in red] There exists a wide variety of tools for SV calling from short reads (Pabinger et al., 2014) but despite ongoing efforts, the discovery of SVs from short-read data remains challenging (English et al., 2015). Studies have estimated that short-read methods suffer from poor sensitivity down to 10% particularly for small SVs shorter than 1 kbp (Chaisson et al., 2015; Huddleston et al., 2016). In contrast to SNPs where discovery and sequence resolution can be performed simultaneously, SVs are discovered mainly indirectly using short paired-end reads. Their alignments are examined for characteristic signatures, such as inconsistently mapping read pairs, split reads and changes in read depth (Alkan et al., 2011). These signatures can only be indirect evidence in favor of certain SV classes but are unable to fully characterize the SV. The main limitation here is that most SVs are simply larger than the short reads. The accurate detection of SVs is, besides their diversity, hampered by their association with repeat regions, biases in the sequencing technology and the additional complexity of diploidy (Carvalho and Lupski, 2016; Huddleston and Eichler, 2016; Willems et al., 2014). To characterize the full spectrum of human genetic variation, long-read sequencing technologies that generate reads with an average length of tens of kilobases show many advantages. The long reads can be mapped with greater accuracy which enables the sequencing of repetitive and low-complexity regions (Chaisson and Tesler, 2012; Loomis et al., 2013). Unlike with short reads, SVs are often spanned by a single long read. This enables the direct detection and full characterization of the SVs. Consequently, several studies confirmed that a substantial number of SVs that are missed by short-read approaches can be identified with long reads (English et al., 2015; Huddleston et al., 2016; Merker et al., 2018). Two commercial long-read sequencing solutions exist to date: single-molecule real-time (SMRT) sequencing by Pacific Biosciences (PacBio) and Nanopore sequencing by Oxford Nanopore Technologies (ONT). Both technologies have the same drawbacks: high error rates of ∼5–15% with dominating Indel errors and still high costs compared to short-read sequencing. Similarly to the detection of SVs from short-read data, the first step toward SV detection from long reads is often the alignment of the reads to a reference genome. Depending on the alignment tool used to produce the alignments, SV detection results can vary substantially as Sedlazeck et al. showed for their tool Sniffles (Sedlazeck et al., 2018a). In that study, SV-spanning long reads were aligned with seven different aligners. Their results showed that one particular aligner, NGMLR, outperformed all the others (including BWA-MEM, Minimap2, LAST and BLASR) on the task (Sedlazeck et al., 2018a). In our study, we analyzed read alignments by NGMLR to detect SVs. In the Supplementary Material, however, we include results for Minimap2 which is an order of magnitude faster than NGMLR (Li, 2018). Read alignments alone