Ratatosk: hybrid error correction of long reads enables accurate variant calling and assembly

Abstract

A major challenge to long read sequencing data is their high error rate of up to 15%. We present Ratatosk, a method to correct long reads with short read data. We demonstrate on 5 human genome trios that Ratatosk reduces the error rate of long reads 6-fold on average with a median error rate as low as 0.22 %. SNP calls in Ratatosk corrected reads are nearly 99 % accurate and indel calls accuracy is increased by up to 37 %. An assembly of Ratatosk corrected reads from an Ashkenazi individual yields a contig N50 of 45 Mbp and less misassemblies than a PacBio HiFi reads assembly.

Structural variant-based pangenome construction has low sensitivity to variability of haplotype-resolved bovine assemblies

“…Correct-then-assemble approaches like Canu 58 can be practical for smaller genomes 59 , but on gigabase-sized mammalian genomes like in Bovinae we observed >20 Tb of peak temporary storage and >25k CPU hours for correcting only 30-fold ONT reads. Even recent reference-guided correction approaches like Ratatosk 60 still needed approximately 15k CPU hours to correct 55-fold ONT reads. Cutting-edge sequencing and bioinformatic improvements 61 , 62 , like the ONT Guppy5 basecaller, will likely assist more efficient assembly, resulting in higher QV and reduced computational load; however, currently the ONT specific requirements might be computationally prohibitive, especially when assembling many samples.…”

Scarless excision of an insertion sequence restores capsule production and virulence in Acinetobacter baumannii

“…The long reads from MinION (Oxford Nanopore Technologies) were demultiplexed and base-called using Guppy v3.2.2 (high accuracy model) and subsequently were adaptor and quality (Q ≤ 8) trimmed using Porechop v0.2.2 ( https://github.com/rrwick/Porechop ) and BBDuk ( https://sourceforge.net/projects/bbmap/ ) with default settings, respectively. The short reads were used to polish the long reads employing Ratatosk v0.7.0 [ 29 ] with default settings. The corrected long reads were then de novo assembled using Flye [ 30 ] v2.9 resulting in circular chromosomal contigs.…”

Structural variant-based pangenome construction has low sensitivity to variability of haplotype-resolved bovine assemblies

“…Correct-then-assemble approaches like Canu (Koren et al, 2017) can be practical for smaller genomes (Wick & Holt, 2021), but on gigabase-sized mammalian genomes like in Bovinae we observed >20 Tb of peak temporary storage and >25k CPU hours for correcting only 30-fold ONT reads. Even recent referenceguided correction approaches like Ratatosk (Holley et al, 2021) still needed approximately 15k CPU hours to correct 55-fold ONT reads. Cutting-edge sequencing and bioinformatic improvements (Baid et al, 2021;Silvestre-Ryan & Holmes, 2021), like the ONT Guppy5 basecaller, will likely assist more efficient assembly, resulting in higher QV and reduced computational load; however, currently the ONT specific requirements might be computationally prohibitive, especially when assembling many samples.…”

“…Flye (Kolmogorov et al, 2019) (version 2.8.3-b1725) assemblies were constructed with "-genome-size=2.7g --nano-corr" from Ratatosk (version 0.4) (Holley et al, 2021) errorcorrected nanopore reads. The nanopore reads were corrected using a reference-guided approach, taking the haplotype-specific hifiasm assembly as the reference.…”