Long-read-based human genomic structural variation detection with cuteSV.
Jiang T, Liu Y, Jiang Y, Li J, Gao Y, Cui Z, Liu Y, Liu B, Wang Y. · 2020
Read Long-read-based human genomic structural variation detection with cuteSV. by Jiang T, Liu Y, Jiang Y, Li J, Gao Y, Cui Z, Liu Y, Liu B, Wang Y. (2020) 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-pmc7477834.txt.
Introduction
Jiang and co-authors present cuteSV, a tool that finds structural changes in the human genome from long reads. Long-read sequencing promises full discovery of such variants. Noisy reads and complex signatures make the task hard.
cuteSV gathers signatures for each type of structural variant with tailored methods. A clustering and refinement step then calls the variants with care. Tests on simulated and real data show higher yields than leading tools. The tool also scales well to large data sets. Speed and sensitivity come together in its design. The software is free online for other teams. The authors frame it as a step toward routine, complete variant maps from long-read data. Such maps can sharpen disease gene hunts.
Catalog record
| CREATOR | Jiang T, Liu Y, Jiang Y, Li J, Gao Y, Cui Z, Liu Y, Liu B, Wang Y. |
|---|---|
| YEAR | 2020 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 76 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:
Long-read-based human genomic structural variation detection with cuteSV By Tao Jiang, Yongzhuang Liu, Yue Jiang, Junyi Li, Yan Gao, Zhe Cui, Yadong Liu, Bo Liu, Yadong Wang Published in: Genome Biology DOI: 10.1186/s13059-020-02107-y ABSTRACT Long-read sequencing is promising for the comprehensive discovery of structural variations (SVs). However, it is still non-trivial to achieve high yields and performance simultaneously due to the complex SV signatures implied by noisy long reads. We propose cuteSV, a sensitive, fast, and scalable long-read-based SV detection approach. cuteSV uses tailored methods to collect the signatures of various types of SVs and employs a clustering-and-refinement method to implement sensitive SV detection. Benchmarks on simulated and real long-read sequencing datasets demonstrate that cuteSV has higher yields and scaling performance than state-of-the-art tools. cuteSV is available at https://github.com/tjiangHIT/cuteSV. BACKGROUND Structural variations (SVs) represent genomic rearrangements such as deletions, insertions, inversions, duplications, and translocations whose sizes are larger than 50 bp [1]. As the largest divergences across human genomes [2], SVs are closely related to human diseases (e.g., inherited diseases [3–5] and cancers [6]), evolution (e.g., gene losses and transposon activity [7, 8]), gene regulations (e.g., rearrangements of transcription factors [9]), and other phenotypes (e.g., mating and intrinsic reproductive isolation [10, 11]). Efforts have been made to develop short-read-based SV calling approaches [12, 13]. Most of them use the methods such as read-depths [14], discordant read-pairs [15], split read alignments [16], local assembly [17], or their combinations [18–20], and they have played important roles in large-scale genomics studies such as 1000 Genomes Project [1]. However, the relatively low read length limits these tools to implement sensitive SV detection [21], and false positives exist as well [22]. With the rapid development of long-read sequencing technologies, such as Pacific Bioscience (PacBio) [23] and Oxford Nanopore Technology (ONT) [24] platforms, long-range spanning information provides the opportunity to more comprehensively detect SVs at a higher resolution [25]. However, novel computational approaches are required to well-handle the high sequencing error rates (typically 5–20%) and large lengths (over 10kbp on average) of the reads [26]. Mainly, two categories of approaches were employed in previous studies, i.e., de novo assembly-based and read alignment-based. De novo assembly-based approaches [27–29] aim at assembling reads to longer genomic sequences (i.e., contigs and/or scaffolds) and discover SVs from the alignments between the assembled sequences and a reference. Such approaches are less influenced by the reference than that of read alignment-based approaches, especially free of read alignment artifacts. However, they are usually computational-intensive and still have some difficulties in the reconstruction of haplotype sequences of large genomes [30], which are also shortcomings to SV calling (a brief discussion is in the “Discussion” section). Read alignment-based approaches directly align reads against the reference and detect SVs by analyzing the alignment results. Such approaches are more cost-effective to computational resources without lack of sensitivity and have been widely used in long read-based SV calling. Several read-alignment-based SV callers have been proposed, such as PB-Honey [31], SMRT-SV [32], Sniffles [33], PBSV (https://github.com/PacificBiosciences/pbsv), and SVIM [34]. They use various methods to find evidence of SVs implied by read alignments, such as the identification of local genomic regions with highly divergent alignments, the local assembly and re-alignment of clipped read parts, and the clustering of SV-spanning signatures [35]. Moreover, state-of-the-art long-read aligners, such as BLASR [36], NGMLR [33], Minimap2 [37], and PBMM2 (https://github.com/PacificBiosciences/pbmm2), were usually employed for read alignment. However, read alignment-based SV calling is still non-trivial. Under the circumstance of high sequencing errors and complicated SVs, the alignment of the reads around SV breakpoints are chimeric and heterogeneous, usually less sensitive and accurate. Therefore, the SV signatures implied by read alignments are highly complicated, and it is difficult to collect and analyze them to implement sensitive detection for various kinds of SVs. Mainly, state-of-the-art tools have the following technical issues to be addressed: (1) overall, the sensitivity is still not satisfying (i.e., a high sequencing coverage is required and/or some SVs are still hard to detect); (2) some approaches (such as rMETL [38], rCANID [39], and npInv [40]) can only detect a subset or a particular class of SVs due to their specific designs; (3) some approaches (such as PBSV and SMRT-SV) are still time-consuming and do not have good scaling performance, which could be not suited to many large datasets; and (4) some approaches (such as SMRT-SV and PB-Honey) only support one type of sequencing data (e.g., for PacBio reads only), as they take the advantage of the characteristics of the data. These drawbacks are still the bottlenecks to the wide use of long-read sequencing data. Herein, we present cuteSV, a versatile read-alignment-based SV detection approach having several beneficial features. (1) cuteSV has better SV detection yields than those of state-of-the-art SV callers. Especially, it has higher sensitivity for low coverage datasets without lack of accuracy. (2) cuteSV supports the datasets produced by mainstream long-read sequencing platforms with various error rates and can discover various types of SVs (including deletions, insertions, duplications, inversions, and translocations). (3) cuteSV has faster or comparable speed to state-of-the-art approaches with lower RAM usage. More importantly, it has outstanding scalability, i.e., enables to achieve almost linear speedup with the number of CPU threads. With these features, cuteSV is suited to large-scale data analysis tasks and has potentials to the cutting-edge genomics studies. RESULTS Overview of cuteSV Using sorted BAM file(s) as input, cuteSV extracts large insertions/deletions and split alignments in aligned reads as SV signatures and clusters and analyzes them to call SVs. The approach has three major steps as follows. Step 1: cuteSV uses multiple signature extraction methods to comprehensively collect the signatures of various types of SVs. Furthermore, the insertions and deletions are heuristically combined to recover the evidence of real SVs from fragile alignments.Step 2: cuteSV uses a specifically designed clustering-and-refinement approach to cluster the chimerically aligned reads in local regions and further refines the clusters to precisely distinguish the SV signatures from heterozygous SVs.Step 3: cuteSV uses several tailored rules to implement SV calling and genotyping based on the refined clusters of SV signatures. A schematic illustration is in Fig. 1, and more details are in the “Materials and methods” section. Fig. 1Schematic illustration of the cuteSV approach. cuteSV uses sorted BAM file as input to detect SVs in 3 major steps. In step 1 (“discovering SV signatures”), cuteSV collects various types of SV signatures comprehensively from inter- and intra-alignments. In step 2 (“clustering of SV signatures”), a heuristic clustering-and-refinement method is employed to sensitively discover accurate SV alleles. In step3 (“SV calling and genotyping”), cuteSV generates the SV callsets and assigns genotypes The major contribution of cuteSV approach is that it uses tailored heuristics to address three difficult technical issues in the alignment-based SV calling. Firstly, due to the scoring systems of read aligners, some large SV events could be divided into several smaller insertions/deletions in a local region. Such cases are usually treated as multiple smaller SVs mistakenly. However, in step 1 of cuteSV, a heuristic method is used to combine insertions/deletions in nearby genomic regions to unbroken signatures of larger SVs. This method not only reduces the errors caused by the fragile read alignments, but also enables to produce more homogenous SV signatures