Reaching the Parts Other Methods

Can’t: Long Reads for Microbial

Genomics and MetagenomicsProf Nick Loman

Institute of Microbiology and Infection

University of Birmingham

COI Disclaimer:

Oxford Nanopore Technologies provided free of charge reagents in support of some of the studies presented here and paid an honorarium for NJL to

speak at a company meeting.

Why bother with long reads?

• De novo assembly– Generate finished references for resequencing projects

– Detection of large scale structural variation

– Plasmid, IS element, reconstruction

– Metagenomics

• Metabarcoding– Full length amplicons (16S, 18S, …)

• Haplotyping– Amplicons

– Strain reconstruction in complex mixtures

Long read technologies

• Single molecule • Synthetic/linked

Long read technologies

• Single molecule

– Contigs

– Noisy (5-15% error)

– Best for assembly

• Synthetic/linked

– Scaffolds

– Very accurate

– Best for haplotyping

10X Genomics

Chromium: 1m GemCodesGemCode: 100k GemCodes

James Hadfield, CRUK

Very low input requirement (1ng)Long fragments needed >50kb5 minute processing time

Patchy coverage within single molecule

Bryan Wee, James Hadfield

Long reads for reference genoems

How long is long enough?

• 7kb reads will

bridge rRNA

operon

• Typically the

longest bacterial

repeat

sequence

Serge Koren, Adam Phillippy

Quick et al. BMJ Open 2014 http://bmjopen.bmj.com/content/4/11/e006278.long

Forensic source tracking of Pseudomonas

aeruginosa in a Burns unit

• ST395 representative sequence by PacBio (P4-C2)

• Resolve to two contigs

With thanks to Lex Nederbragt

SNPs Indels Mapped

PAO1Reference

23 4 77%

PacBioReference

40 5 97%

Choice of reference informs

phylogenetic resolution

Bed 11

Environment

Water

Patient

ΔoprD

ΔlysR-like regulator

Rapid emergence of meropenem resistance

Long reads for metagenomics

Noisy reads for metagenomics

• PacBio – Even HMP mock community

Courtesy of Pacific Biosciences

Long read technologies

• Oxford Nanopore MinION • Portable, fieldable

• Real-time

• Low/no instrument cost

• Reads up to 1Mb

• 1-10Gb per run

• Homopolymeric tract errors

• 1D: 15% error

• 1D^2: 5% error

Nanopore WGS Consortium, https://github.com/nanopore-wgs-consortium/NA12878

R9.4 chemistry enables higher yields

In field whole-genome sequencing

MRSAS. pyogenes

900Mb 1D rapid

P. aeruginosa

E. faecium

Bandage by Ryan Wick

4.4Gb 1D ligation

Whale watching with nanopore

• Sambrook

• Excess of DNA (>10ug)

• V. careful pipetting

• Aim <1 transposase cut per moleculePicture from Phil Zuzerte

Total bases: 5,014,576,373 (5Gb) Number of reads: 150,604N50: 63,747Mean: 33,296.44

http://lab.loman.net/2017/03/09/ultrareads-for-nanopore/

Whale watching: E. coli

1113805 916705 790987 778219 771232 671130 646480 629747 614903 603565

Whale watching: E. coli

“Squiggles are vanity, alignments are sanity!

E. coli: genome assembly in 8 reads

Read Length Ref start Ref end Time (m)

1 876991 4398844 634183 32.48

2 696402 470003 1166405 25.79

3 799047 1137438 1936485 29.59

4 642071 1759431 2401502 23.78

5 826662 2106227 2932889 30.61

6 883962 2699626 3583588 32.73

7 825191 3285196 4110387 30.56

8 463341 3995967 4459308 17.16

miniasm

N50 4MbTime: 1.5s (1 CPU)

1x coverage!

PFGE: The future

What’s coming next?

• Combined single molecule and linked read

hybrid metagenomic assemblies?

• New methods for long fragment extraction from

microbiome studies

• Hi-C to link chromosomes w/ plasmids, phage

• Much higher outputs

Conclusions

• Long reads and synthetic long reads promising for whole-genome reconstruction from metagenomics

• Synthetic long reads (10X and/or hybrid) will be useful for detecting low abundance members of population

• Issues:– Very high input requirements may be challenging for certain

applications, e.g. diagnostics from clinical samples

– Extraction of high molecular weight DNA from mixed samples tricky; no one size fits all solution

– Cost remains prohibitive for very deep sequencing

Acknowledgements

– Josh Quick (Birmingham)

– 10X:

• Pablo Fuentes-Utrilla (Birmingham)

• Bryan Wee, Ross Fitzgerald (Roslin)

• James Hadfield (CRUK)

– Nanopore human genome sequencing consortium

– Pseudomonas

• Nicola Cumley, Beryl Oppenheim

Synthetic long reads

• Moleculo

– Long range amplification

– “Nextera on steroids”

– 500ng input

– Size selection required

http://www.illumina.com/products/truseq-synthetic-long-read-kit.ilmn