Sequencing of human genomes with nanopore technology

Rory Bowden ,David Buck ,Michael A Simpson ,Hannah Roberts ,Alistair T Pagnamenta ,Andreas Heger ,Silvia Salatino ,Fatima Dhalla ,Jenny C Taylor ,Duncan Parkes ,R W Davies ,Ip Clc ,Niko Popitsch ,Mariateresa De Cesare ,Smita Y Patel ,Colin Freeman ,Helen Lockstone ,Gerton Lunter ,Laura Oikkonen ,Peter Donnelly

doi:10.1038/s41467-019-09637-5

Abstract

Whole-genome sequencing (WGS) is becoming widely used in clinical medicine in diagnostic contexts and to inform treatment choice. Here we evaluate the potential of the Oxford Nanopore Technologies (ONT) MinION long-read sequencer for routine WGS by sequencing the reference sample NA12878 and the genome of an individual with ataxia-pancytopenia syndrome and severe immune dysregulation. We develop and apply a novel reference panel-free analytical method to infer and then exploit phase information which improves single-nucleotide variant (SNV) calling performance from otherwise modest levels. In the clinical sample, we identify and directly phase two non-synonymous de novo variants in SAMD9L, (OMIM #159550) inferring that they lie on the same paternal haplotype. Whilst consensus SNV-calling error rates from ONT data remain substantially higher than those from short-read methods, we demonstrate the substantial benefits of analytical innovation. Ongoing improvements to base-calling and SNV-calling methodology must continue for nanopore sequencing to establish itself as a primary method for clinical WGS.

Highlights

Whole-genome sequencing (WGS) is becoming widely used in clinical medicine in diagnostic contexts and to inform treatment choice
Initial reports of nanopore human genome sequencing have focused on the benefits of Oxford Nanopore’s long reads in achieving a highly contiguous assembly[16] or in identifying structural variation in patient samples[17]. This focus is unsurprising, considering the potential difficulties presented by the modest per-base accuracy of Oxford Nanopore reads in attaining the low genome-wide error rates for single-nucleotide variant (SNV) calling that will be required for most clinical applications of genome sequencing using a single technology
We evaluated the performance of singlenucleotide variant (SNV) calling in NA12878, using the multiplatform Genomes in a Bottle (GIAB) variant calls as a gold standard truth data set[18]

Summary

Introduction

Whole-genome sequencing (WGS) is becoming widely used in clinical medicine in diagnostic contexts and to inform treatment choice. The sequencing-by-synthesis methodology initially developed and commercialised by Solexa, and progressively improved in both accuracy and throughput by Illumina, has transformed the study of the human genome for research and clinical applications[3,4] Despite such advances, short-read lengths restrict the insight that can be derived from sequencing of an individual genome by limiting the resolution of repetitive regions, complex structural variation, and haplotype phase. Initial reports of nanopore human genome sequencing have focused on the benefits of Oxford Nanopore’s long reads in achieving a highly contiguous assembly[16] or in identifying structural variation in patient samples[17] This focus is unsurprising, considering the potential difficulties presented by the modest per-base accuracy of Oxford Nanopore reads in attaining the low genome-wide error rates for single-nucleotide variant (SNV) calling that will be required for most clinical applications of genome sequencing using a single technology. We sequenced DNA from an individual with ataxia-pancytopenia syndrome accompanied by severe immune dysregulation in order to fully resolve a question relating to the phasing of two de novo protein-coding variants that is relevant for a complete molecular genetic diagnosis

Methods

Results

Conclusion