Vulcan: Improved long-read mapping and structural variant calling via dual-mode alignment

Abstract

Background Long-read sequencing has enabled unprecedented surveys of structural variation across the entire human genome. To maximize the potential of long-read sequencing in this context, novel mapping methods have emerged that have primarily focused on either speed or accuracy. Various heuristics and scoring schemas have been implemented in widely used read mappers (minimap2 and NGMLR) to optimize for speed or accuracy, which have variable performance across different genomic regions and for specific structural variants. Our hypothesis is that constraining read mapping to the use of a single gap penalty across distinct mutational hot spots reduces read alignment accuracy and impedes structural variant detection. Findings We tested our hypothesis by implementing a read-mapping pipeline called Vulcan that uses two distinct gap penalty modes, which we refer to as dual-mode alignment. The high-level idea is that Vulcan leverages the computed normalized edit distance of the mapped reads via minimap2 to identify poorly aligned reads and realigns them using the more accurate yet computationally more expensive long-read mapper (NGMLR). In support of our hypothesis, we show that Vulcan improves the alignments for Oxford Nanopore Technology long reads for both simulated and real datasets. These improvements, in turn, lead to improved accuracy for structural variant calling performance on human genome datasets compared to either of the read-mapping methods alone. Conclusions Vulcan is the first long-read mapping framework that combines two distinct gap penalty modes for improved structural variant recall and precision. Vulcan is open-source and available under the MIT License at https://gitlab.com/treangenlab/vulcan.

The De Novo Genome Assembly of Olea europaea subsp. cuspidate, a Widely Distributed Olive Close Relative

“…Although the “assembly-to-assembly” approach has been successfully used to identify SVs in other species ( Chen et al, 2022b ; Goel et al, 2019 ), we failed to obtain results from this method, probably due to the known phenomenon of higher rearrangements in plants than in animals. We further identified structural variations (SVs) using Sniffles V2.0.3 ( Sedlazeck et al, 2018 ) and a dual-alignment strategy implemented in Vulcan ( Fu et al, 2021 ). Vulcan explores the advantages of two efficient mappers, Minimap2 ( Li, 2018 ) and NGMLR ( Sedlazeck et al, 2018 ), to improve the accuracy and efficiency of mapping.…”

“…We further conducted the SV identification based on comparing OC long reads to OS and OE references with a dual-mode alignment strategy. In detail, the reads were mapped to a reference with two commonly used mappers, Minimap2 and NGMLR, which are integrated in a software named Vulcan ( Fu et al, 2021 ). Minimap2 is a highly fast long-read mapper, implementing a time-efficient alignment approach involving a two-piece affine gap model and a faster chaining process ( Li, 2018 ).…”

Benchmarking Oxford Nanopore Read Alignment-Based Structural Variant Detection Tools in Crop Plant Genomes

“…The SV callers included in the study were selected using several criteria: (1) citation count (used as a proxy for popularity in the research community), (2) publication date and maintenance status (excluding older tools that were no longer maintained), (3) ability to detect both insertion and deletion SVs from ONT data. The benchmarking approach involved four long-read aligners, including minimap2 (Li, 2018), NGMLR (Sedlazeck et al, 2018a), lra (Ren and Chaisson, 2021), and Vulcan (Fu et al, 2021) as well as five SV calling software namely Sniffles (v2) (Sedlazeck et al, 2018b), NanoVar (Tham et al, 2019), SVIM (Heller and Vingron, 2019), cuteSV (Jiang et al, 2020), and dysgu (Cleal and Baird, 2022). All aligners and SV caller versions are provided in detail in ( Table S1 ).…”

“…Several algorithms were developed for SV discovery from long-reads including Sniffles (Sedlazeck et al, 2018b), NanoVar (Tham et al, 2019), SVIM (Heller and Vingron, 2019), cuteSV (Jiang et al, 2020), and dysgu (Cleal and Baird, 2022), which have been comprehensively reviewed recently (Mahmoud et al, 2019; Yuan et al, 2021). Additionally, several long-read aligners are available such as minimap2 (Li, 2018), NGMLR (Sedlazeck et al, 2018a), Vulcan (Fu et al, 2021), and lra (Ren and Chaisson, 2021). Considering the continued development and improvement in read-alignment and SV detection algorithms and multitude of their possible combinations, their combined performances in SV detection demand realistic and up-to-date benchmarks to guide the selection of SV discovery tools.…”

A survey of mapping algorithms in the long-reads era

“…While this survey covers the genomic mapping aspects, other important contributions have dealt with adapted procedures in the case of long-read RNA mapping [53, 65, 50, 74], and structural variant identification [68, 48, 24, 73], or other specialized problems [55]. Other related research focused on read-to-read overlap detection [20, 75] 2 , or alignment-free/pseudo-mapping approaches [33, 13].…”