RaGOO: fast and accurate reference-guided scaffolding of draft genomes.
Alonge M, Soyk S, Ramakrishnan S, Wang X, Goodwin S, Sedlazeck FJ, Lippman ZB, Schatz MC. · 2019
Read RaGOO: fast and accurate reference-guided scaffolding of draft genomes. by Alonge M, Soyk S, Ramakrishnan S, Wang X, Goodwin S, Sedlazeck FJ, Lippman ZB, Schatz MC. (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-pmc6816165.txt.
Introduction
Alonge and co-authors present a software tool in Genome Biology. Genome assembly often yields many short pieces called contigs. To read a genome well, those pieces must be ordered into full chromosomes. RaGOO does that job by aligning contigs to a finished reference genome. It uses the fast aligner Minimap2.
The authors show RaGOO ordering three draft tomato genomes, including the widely used M82 line, in minutes. After building the chromosome-scale sequences, the tool also flags structural variants, even across gaps. They then applied it to 103 strains of the small plant Arabidopsis. That scan showed that genes for defence were among the most variable in the group.
Catalog record
| CREATOR | Alonge M, Soyk S, Ramakrishnan S, Wang X, Goodwin S, Sedlazeck FJ, Lippman ZB, Schatz MC. |
|---|---|
| YEAR | 2019 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 74 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:
RaGOO: fast and accurate reference-guided scaffolding of draft genomes By Michael Alonge, Sebastian Soyk, Srividya Ramakrishnan, Xingang Wang, Sara Goodwin, Fritz J. Sedlazeck, Zachary B. Lippman, Michael C. Schatz Published in: Genome Biology DOI: 10.1186/s13059-019-1829-6 ABSTRACT We present RaGOO, a reference-guided contig ordering and orienting tool that leverages the speed and sensitivity of Minimap2 to accurately achieve chromosome-scale assemblies in minutes. After the pseudomolecules are constructed, RaGOO identifies structural variants, including those spanning sequencing gaps. We show that RaGOO accurately orders and orients 3 de novo tomato genome assemblies, including the widely used M82 reference cultivar. We then demonstrate the scalability and utility of RaGOO with a pan-genome analysis of 103 Arabidopsis thaliana accessions by examining the structural variants detected in the newly assembled pseudomolecules. RaGOO is available open source at https://github.com/malonge/RaGOO. BACKGROUND Long-read single-molecule sequencing technologies commercialized by Oxford Nanopore Technologies (ONT) and Pacific Biosciences (PacBio) have facilitated a resurgence of high-quality de novo eukaryotic genome assemblies [1]. Assemblies using these technologies in a variety of plant and animal species have consistently reported contig N50s over 1 Mbp, while also reconstructing higher percentages of target genomes, including repetitive sequences [2, 3]. Current long-read sequencers are now able to produce over one terabase of long reads per week, presenting the opportunity for detailed pan-genome analysis of unprecedented scale. Such analyses can include structural variations that are notoriously difficult to detect using short-read sequencing. However, lagging behind the current speed and cost of generating long-read sequencing data are genome assemblers, which are still unable to resolve complex repeats and related structural variants that are widespread in eukaryotic genomes. Thus, there is a need for simplified and faster approaches to scaffold fragmented genome assemblies into chromosome-scale pseudomolecules. Two common approaches have been used to achieve chromosome-scale assemblies, namely, reference-free (de novo) and reference-guided approaches. One popular reference-free scaffolding approach is to anchor genome assembly contigs to some variety of genomic map [4], such as an optical, physical, or linkage map [5]. This process involves aligning the genomic map to a sequence assembly and scaffolding contigs according to the chromosomal structure indicated in the map. However, contigs not implicated in any alignments will fail to be scaffolded, which can result in incomplete scaffolding. Furthermore, acquiring a genomic map can be expensive, time-consuming, or otherwise intractable depending on the species and the type of map. Another reference-free method for pseudomolecule construction involves the use of long-range genomic information to scaffold assembled contigs. This includes a large class of technologies such as mate-pair sequencing, Bacterial Artificial Chromosomes (BACs), Linked Reads and chromatin conformation capture such as Hi-C [6–8]. In particular, Hi-C has recently been shown to be a practical and effective resource for chromosome-scale scaffolding [9–11]. Paired-end Hi-C sequencing reads are aligned to the assembly, and mates which align to different contigs (Hi-C links) are recorded. According to the relative density of such Hi-C links between pairs of contigs, contigs can be ordered and oriented into larger scaffolds, potentially forming chromosome-length pseudomolecules. Also, because misassemblies may be observed by visualizing Hi-C alignments, Hi-C can be used for validation and manual correction of misassemblies [12]. Though Hi-C has been widely adopted, there remain challenges that can impede the ability to form accurate chromosome-scale pseudomolecules with Hi-C alone. Principally, Hi-C data are noisy, and Hi-C-based scaffolders are prone to producing structurally inaccurate scaffolds [13]. Also, because this process relies on the alignment of short Hi-C sequencing reads to the draft assembly, small and repetitive contigs with little or conflicting Hi-C link information often fail to be accurately scaffolded. Finally, the analysis requires deep sequencing coverage and therefore can be expensive and compute-intensive. Aside from reference-free approaches, there are also a few tools available for reference-guided scaffolding [14]. For example, Chromosomer and MUMmer’s “show-tiling” utility leverage pairwise alignments to a reference genome for contig scaffolding and have been used to scaffold eukaryotic genomes [15–18]. RACA is similar, though it also requires paired-end sequencing data to aid scaffolding [19]. Finally, tools such as GOS-ASM and Ragout2 employ multiple sequence aligners to reconcile multiple, potentially diverse contig sets [20, 21]. Though reference-guided scaffolding may introduce erroneous reference bias, it is often substantially faster and less expensive than acquiring the resources for the reference-free methods outlined above. However, current tools for reference-guided scaffolding of eukaryotic genomes have notable shortcomings. Firstly, these tools depend on slower DNA aligners such as BLAST and Nucmer and accordingly require long compute times of several hours to several days for mammalian-sized genomes [22]. This is especially pronounced in tools like Ragout2 that use multiple sequence aligners, such as Cactus, that can require hundreds of CPU hours for large eukaryotic genomes [23]. These aligners are also not robust to repetitive and/or gapped alignments resulting in a significant portion of contigs being unassigned in pseudomolecules. Finally, many of these methods do not internally offer the ability to correct large-scale misassemblies frequently present in draft assemblies of eukaryotic genomes nor report any metrics on conflicts due to true biological differences in the genomes. Here, we introduce RaGOO, an open-source method which utilizes Minimap2 [24] alignments to a closely related reference genome to quickly cluster, order, and orient genome assembly contigs into pseudomolecules. RaGOO also provides the option to correct apparent chimeric contigs prior to pseudomolecule construction. Finally, structural variants (SVs), including those spanning gap sequence, are identified using an optimized and integrated version of Assemblytics [25], thus enabling rapid pan-genome SV analysis of many genomes at once. This is especially important for detecting large insertions and other complex structural variations that are difficult to detect using read mapping approaches. We first demonstrate the speed and accuracy of RaGOO scaffolding with simulated data of increasing complexity and show that it outperforms 2 popular alternative methods. We next show the utility of RaGOO by creating high-quality chromosome-scale reference genomes for 3 distinct wild and domesticated genotypes of the model crop tomato using a combination of short and long-read sequencing. Finally, we demonstrate the scalability of RaGOO by ordering and orienting 103 draft Arabidopsis thaliana genomes and comparing structural variants across the pan-genome. This uncovers a large number of defense response genes that are highly variable. RESULTS Reference-guided contig ordering and orientation with RaGOO RaGOO is a fast and reliable reference-guided scaffolding method, implemented as an open-source python command-line utility, that orders and orients genome assembly contigs according to Minimap2 alignments to a single reference genome (Fig. 1) [26]. RaGOO’s primary goal is to utilize the large-scale structure of a reference genome to organize assembly contigs, analogous to how a genetic map is used. Therefore, under default settings, RaGOO does not alter or mutate any input assembly sequence but rather arranges them and places gaps for padding between contigs. Additionally, users have the option to break input contigs at points of potential misassembly indicated by discordant alignments to the reference genome. However, these breaks will only fragment the assembly and do not add or remove any sequence content. RaGOO can optionally avoid breaking chimeric intervals at loci within genomic coordinates specified by a gff3 file, so as to avoid disrupting gene models identified in the de novo assembly. Fig. 1The RaGOO pipeline. a Contigs are aligned to the reference genome with Minimap2