Nanopore sequencing and the Shasta toolkit enable efficient de novo assembly of eleven human genomes

Kishwar Shafin,Karen H Miga,Kristof Tigyi,Hugh E Olsen,Kelvin Liu ,P Carnevali ,Fritz J Sedlazeck,Edward Green ,Justin M Zook,Melanie Sorensen,Duncan Kilburn,Erik Garrison,Marina Haukness,Tobias Marschall,Simon Mayes,Adam M Phillippy ,Mitchell R Vollger,Mark Akeson,Colleen M Bosworth ,Trevor Pesout,Miten Jain,Sofie R Salama ,Katherine M Munson ,Sergey Koren,David Haussler,Nicholas Maurer,Vǎnia Costa ,Joel Armstrong,Ryan Lorig-Roach,Evan E Eichler,Jean Monlong,Benedict Paten

doi:10.1038/s41587-020-0503-6

Abstract

De novo assembly of a human genome using nanopore long-read sequences has been reported, but it used more than 150,000 CPU hours and weeks of wall-clock time. To enable rapid human genome assembly, we present Shasta, a de novo long-read assembler, and polishing algorithms named MarginPolish and HELEN. Using a single PromethION nanopore sequencer and our toolkit, we assembled 11 highly contiguous human genomes de novo in 9 d. We achieved roughly 63× coverage, 42-kb read N50 values and 6.5× coverage in reads >100 kb using three flow cells per sample. Shasta produced a complete haploid human genome assembly in under 6 h on a single commercial compute node. MarginPolish and HELEN polished haploid assemblies to more than 99.9% identity (Phred quality score QV = 30) with nanopore reads alone. Addition of proximity-ligation sequencing enabled near chromosome-level scaffolds for all 11 genomes. We compare our assembly performance to existing methods for diploid, haploid and trio-binned human samples and report superior accuracy and speed.

Highlights

Reference-based methods such as GATK1 can infer human variations from short-read sequences, but the results only cover ~90% of the reference human genome assembly[2,3]
A total of two nuclease flushes were performed per flow cell, and each flow cell received a total of three sequencing libraries
For the HG00733 and CHM13 samples we found that Shasta assemblies polished with MarginPolish and homopolymer encoded long-read error-corrector for Nanopore (HELEN) contained most human protein coding genes, having, respectively, an identified ortholog for 99.23% (152 missing) and 99.11% (175 missing) of these genes

Summary

Introduction

Reference-based methods such as GATK1 can infer human variations from short-read sequences, but the results only cover ~90% of the reference human genome assembly[2,3]. These methods are accurate with respect to single-nucleotide variants and short insertions and deletions (indels) in this mappable portion of the reference genome[4]. In addition to increasingly being used in reference guided methods[2,14,15,16], long-read sequences can generate highly contiguous de novo genome assemblies[17]. As commercialized by Oxford Nanopore Technologies (ONT), is useful for de novo genome assembly because it can produce high yields of very long 100+ kilobase (kb) reads[18]. We use a combination of nanopore and proximity-ligation (HiC) sequencing[9] and our toolkit, and we report improvements in human genome sequencing coupled with reduced time, labor and cost

Methods

Results

Conclusion