YaHS: yet another Hi-C scaffolding tool.
Zhou C, McCarthy SA, Durbin R. · 2023
Read YaHS: yet another Hi-C scaffolding tool. by Zhou C, McCarthy SA, Durbin R. (2023) 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-pmc9848053.txt.
Introduction
Zhou and co-authors present YaHS, a command-line tool that builds chromosome-scale scaffolds from Hi-C data. Genome projects now assemble many species, yet long reads alone rarely yield full chromosomes. Hi-C contact data can order and join contigs into large scaffolds.
YaHS runs from a single-line command and needs only an assembly file and an alignment file. It returns results in several formats and aims for speed, accuracy, and contiguity. Tests on a simulated human assembly placed over 92 percent of sequence into 25 major scaffolds. Its scaffold N50 far exceeded two rival tools in that test. The tool is written in C under the MIT License.
Catalog record
| CREATOR | Zhou C, McCarthy SA, Durbin R. |
|---|---|
| YEAR | 2023 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 17 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:
YaHS: yet another Hi-C scaffolding tool By Chenxi Zhou, Shane A McCarthy, Richard Durbin, Can Alkan Published in: Bioinformatics DOI: 10.1093/bioinformatics/btac808 ABSTRACT Abstract Summary We present YaHS, a user-friendly command-line tool for the construction of chromosome-scale scaffolds from Hi-C data. It can be run with a single-line command, requires minimal input from users (an assembly file and an alignment file) which is compatible with similar tools and provides assembly results in multiple formats, thereby enabling rapid, robust and scalable construction of high-quality genome assemblies with high accuracy and contiguity. Availability and implementation YaHS is implemented in C and licensed under the MIT License. The source code, documentation and tutorial are available at https://github.com/sanger-tol/yahs. Supplementary information Supplementary data are available at Bioinformatics online. 1 INTRODUCTION The rapid revolution of long-read, single-molecule DNA sequencing technologies in read length, base accuracy and per-base cost is driving a golden age for de novo genome assembly. Multiple genome sequencing projects have been launched in the past few years, such as the Earth Biogenome Project (Lewin et al., 2018), the Vertebrate Genomes Project (Rhie et al., 2021) and the Darwin Tree of Life Project (DToL, Blaxter et al., 2022) aiming to assemble high quality, chromosome-scale genomes for many thousands of species across a range of genome sizes, complexity and ploidy. Despite the technological advances, the assembly of reference-quality genomes with long-read sequencing data alone remains elusive (Amarasinghe et al., 2020). Long-distance linkage information, such as physical maps, genetic maps, optical maps and Hi-C contact maps, is often used to construct chromosome-scale scaffolds from contigs. Hi-C is a sequencing-based proximity ligation assay that provides contact information between pairs of loci, originally designed to study the 3D structure of the genome inside a cell nucleus (Lieberman-Aiden et al., 2009). Since the contact frequency between loci pairs strongly correlates with separation on the genome, Hi-C has rapidly gained popularity as an economical method for generating chromosome-scale scaffolds (Burton et al., 2013; Dudchenko et al., 2017). Several scaffolding tools have been developed for the construction of chromosome-scale assembly with Hi-C data including LACHESIS (Burton et al., 2013), HiRise (Putnam et al., 2016), 3D-DNA (Dudchenko et al., 2017), ALLHiC (Zhang et al., 2019), SALSA2 (Ghurye et al., 2019) and pin_hic (Guan et al., 2021). Each of these has its own limitations, and the results are affected by various factors such as genome complexity and repeat content, Hi-C library preparation and sequencing coverage (Ghurye et al., 2019; Guan et al., 2021; Kadota et al., 2020). In this article, we introduce YaHS, another scaffolding tool which constructs chromosome-scale scaffolds utilizing Hi-C data. YaHS follows a standard framework of Hi-C scaffolding pipelines: map Hi-C reads to input contigs, break contigs where necessary to correct assembly errors, build a contact matrix, construct and prune a scaffolding graph and finally output scaffolds. The tool assumes the input contigs are derived from a single haplotype. If the genome assembly contains haplotypic duplications, they should be removed first. The core idea behind YaHS, which distinguishes it from other Hi-C scaffolding tools, is a novel method for building the contact matrix (Supplementary Methods). This method enables more accurate inferences of contig joins. In comparisons with SALSA2 and pin_hic, two recently published Hi-C scaffolding tools, applied to both simulated and real data, YaHS generated genome assemblies of higher accuracy and contiguity, and was more robust to assembly errors. 2 RESULTS 2.1 Overview The scaffolding process starts with mapping Hi-C reads to the input contigs, which falls outside of the scope of YaHS. The Arima Genomics mapping pipeline was employed in this study which consists of four major steps: read mapping in single-end read mode, read filtering for chimeric joins across ligation junctions, read pairing and PCR duplicate removal (https://github.com/ArimaGenomics/mapping_pipeline). YaHS takes the alignment file (either in BED format or BAM format) to first optionally break contigs at positions lacking Hi-C coverage which are potential assembly errors. Scaffolding then proceeds in multiple rounds. In each round YaHS builds a contact matrix by splitting each contig into chunks of a certain size (i.e. resolution) and assigns Hi-C contact signals into cells of chunk pairs. Here, we refer to cells within contigs as intra-cells and between contigs as inter-cells. The Hi-C contact frequencies are counted for each cell. To calculate the joining score of a pair of contigs, the contact frequencies of the inter-cells between them are normalized by expected values which are estimated by the medians of the intra-cells at the same separations and then used to calculate a weighted sum. The fundamental idea is that the inter-cells of neighbouring contigs on a scaffold should have similar contact frequencies to intra-cells at the same separations. The joining scores of a contig pair are calculated in all four possible orientations and the one with the largest score is selected as the joining orientation. YaHS also optionally takes account of the restriction enzymes used in the Hi-C library, and if so, the cell contact frequencies are normalized first by the corresponding number of cutting sites. YaHS next builds a scaffolding graph with contigs as nodes and contig joins as edges which are weighted by the joining scores calculated in the previous step. The graph is simplified by a series of operations including filtering low-score edges, trimming tips, trimming blunt ends, solving repeats, removing transitive edges, removing bubbles, resolving ambiguous orientations, trimming weak edges and removing ambiguous edges. Finally, the graph is traversed to assemble scaffolds along contiguous paths. A second step of assembly error correction is optionally performed to break scaffolds at positions of contig joins without sufficient Hi-C coverage. YaHS runs a hierarchical joining process with multiple rounds of scaffolding at decreasing resolutions (increasing chunk sizes). In each round, the scaffolds generated in the previous round are used as the input except for the first round when the contigs are used. See Supplementary Methods for further details. 2.2 On simulated human genome assemblies We randomly split the Telomere-to-Telomere (T2T) human genome assembly (T2T-CHM13, Nurk et al., 2022) into 100 kb to 1 Mb chunks and ended up with a simulated assembly of 5483 contigs with an N50 of 715 kb. The T2T Arima Hi-C data were downloaded from NCBI (accession SRX10230901–SRX10230903) and mapped to the simulated assembly for the reconstruction of the genome with YaHS, SALSA2 and pin_hic. YaHS assembled over 92% sequences into 25 major scaffolds (>35 Mb). The N50 and N90 were 132.6 Mb (L50 = 9) and 36.5 Mb (L90 = 23), respectively (Fig. 1A). In contrast, the N50 and N90 of SALSA2 assembly were 43.9 Mb (L50 = 22) and 3.3 Mb (L90 = 103), respectively (Fig. 1B), and the N50 and N90 of pin_hic assembly were 51.5 Mb (L50 = 19) and 8.5 Mb (L90 = 68), respectively (Fig. 1C). We used QUAST-LG (Mikheenko et al., 2018) to map the scaffolds to the T2T-CHM13 genome assembly for quality assessment. Three types of misassemblies were considered: relocations (gaps on scaffolds), inversions (misorientated contigs) and translocations (misjoins of inter-chromosomal contigs). The numbers of relocations and inversions reported by QUAST-LG for YaHS, SALSA2 and pin_hic assemblies were 40 and 21, 262 and 55, and 171 and 115, respectively. No translocation was reported for any of the three assemblies. [Figure: Hi-C contact