Comparison of De Novo Assembly Strategies for Bacterial Genomes

Abstract

(1) Background: Short-read sequencing allows for the rapid and accurate analysis of the whole bacterial genome but does not usually enable complete genome assembly. Long-read sequencing greatly assists with the resolution of complex bacterial genomes, particularly when combined with short-read Illumina data. However, it is not clear how different assembly strategies affect genomic accuracy, completeness, and protein prediction. (2) Methods: we compare different assembly strategies for Haemophilus parasuis, which causes Glässer’s disease, characterized by fibrinous polyserositis and arthritis, in swine by using Illumina sequencing and long reads from the sequencing platforms of either Oxford Nanopore Technologies (ONT) or SMRT Pacific Biosciences (PacBio). (3) Results: Assembly with either PacBio or ONT reads, followed by polishing with Illumina reads, facilitated high-quality genome reconstruction and was superior to the long-read-only assembly and hybrid-assembly strategies when evaluated in terms of accuracy and completeness. An equally excellent method was correction with Homopolish after the ONT-only assembly, which had the advantage of avoiding hybrid sequencing with Illumina. Furthermore, by aligning transcripts to assembled genomes and their predicted CDSs, the sequencing errors of the ONT assembly were mainly indels that were generated when homopolymer regions were sequenced, thus critically affecting protein prediction. Polishing can fill indels and correct mistakes. (4) Conclusions: The assembly of bacterial genomes can be directly achieved by using long-read sequencing techniques. To maximize assembly accuracy, it is essential to polish the assembly with homologous sequences of related genomes or sequencing data from short-read technology.

Nanopore Sequencing for De Novo Bacterial Genome Assembly and Search for Single-Nucleotide Polymorphism

“…However, in some cases a relevant reference genome is not known. In this case, a de novo genome assembly is required, involving the construction of a complete nucleotide sequence without a reference [ 1 ]. This task is extremely difficult if using next-generation sequencing (NGS), because the typical lengths of reads are hundreds of nucleotides long [ 2 ].…”

Nanopore-only assemblies for genomic surveillance of the global priority drug-resistant pathogen, Klebsiella pneumoniae

“…Short-read-first hybrid assemblies were generated using Unicycler v0.4.9b, which starts by building a short-read assembly graph with SPAdes v3.14.0, then uses the corresponding long reads (HAC basecalled, in this case) to scaffold the genome, and finally runs Pilon v1.23 [58] in an attempt to fill gaps, correct bases and fix misassemblies using the short reads [26, 54] (HAC basecalled reads were used for the short-read-first assemblies as these were generated before the SUP basecalling model became available [50]). To generate long-read-first assemblies [59], we used Flye v2.8-1 to produce ONT-only assemblies for the set of reads basecalled with the HAC and SUP-accuracy basecalling models, followed by long-read polishing with Medaka [30] to repair any residual errors using ONT long reads, then finally short-read polishing using Illumina reads and Pilon v1.24 (following the recommendations noted at [60]). Thus, altogether, we produced ten assemblies per sample, encompassing reads derived from the three separate basecalling models ( Figure 1 ).…”

Correcting Modification-Mediated Errors in Nanopore Sequencing by Nucleotide Demodification and in silico Correction

“…Thanks to several machine-learning algorithms, these errors have been significantly reduced by read-based (e.g., Medaka) or reference-based (e.g., Homopolish) polishing methods [5]. These algorithmic advances have produced high-quality ONT genomes sufficient for downstream analysis (e.g., >Q50) [3, 6].…”