1. Ion Torrent and Minion
Relatively low cost ‘next generation’ sequencing
Wendy Smith
School of Computing Science,
Alan Ward
Newcastle University, UK

2. Overview Projects Sequencing systems Ion Torrent - introduction
- steps involved in producing a sequenced genome
- costs
- results!

3. What is Next Generation Sequencing?
Ion Torrent is CURRENT generation sequencing
Minion MAY be next generation?
Like Sanger, most currently available NGS systems extend a primed template to determine a DNA sequence
Shorter reads – in bp range versus 1000bp for Sanger
BUT perform VERY LARGE numbers (million or more) of ‘short’ reads in parallel in single run to compile a database
= increased coverage – sufficiently large to cover genomes
– a high throughput approach
PacBio/Oxford Nanpore (Minion/Gridion) are single molecule
and long reads
Oxford nanopore is not sequencing by synthesis
Already in use in rapidly increasing numbers of labs

4. Major players - the big three
Company Sequencing system
Illumina Illumina (= Solexa) (market leader)
ABI ‘SOLiD’
Roche ‘454’ sequencing
Advantages: Established commercially since 2008 – well tested
Limitations: COST
Initial outlay for machine is high 150K and cost for a single run is approx 40K so need lots of samples to run at same time- commercial use

5. More affordable recent additions
Typically quarter of the cost of the ‘big three’
Company Sequencing system
Life Technologies IonTorrent - available since end of 2010
Illumina MiSeq
Oxford Nanopore Nanopore ‘trialled’ in selected labs, but only just emerging
Pacific Biosciences PacBio commercially (2012)
Various others still under development

6. IonTorrent – overall approach (similarities to ‘big three’)
DNA fragments immobilised on tiny beads (called ‘Ion Sphere Particles’ or ISPs) – aiming for a SINGLE fragment per bead
Fragments (templates) amplified by PCR – multiple copies of same template molecule on any particular bead
Beads distributed amongst > million individual wells on a reaction ‘chip’ (aiming for one bead per well) – each well a separate ‘reaction chamber’ with separate ‘sensor’
Detect incorporation of specific dNTPs by DNA polymerase
Individual dNTPs are passed over chip one at a time and on each ‘pass’ the sensors identify the wells where a particular dNTP has been incorporated.

7. How does it work?
When a nucleotide is incorporated into a strand of DNA by DNA polymerase, a hydrogen ion is released as a byproduct. The Ion Torrent essentially detects this released H+ to identify when a nucleotide has been incorporated.

8. Ion sequencing chip
Well over a million individual wells – each acting as a separate reaction chamber

9. The chip: an individual well
Each well holds a different DNA template. Beneath the wells is an ion-sensitive layer and beneath that an Ion sensor. This detects the H+ ion released whenever a dNTP is incorporated into the extending sequence in this individual well and hence ‘calls’ the relevant base determined by the template in that particular well. – more on this a bit later.
captures voltage measurements. The total number of independent measurements, or sequence reads, is a function of the number of sensors and fabricated wells that a chip contains –

10. Overview of sequencing workflow
Schematic representation of the ion torrent sequencing workflow. A sequence library is produces by generating DNA fragments flanked by the Ion Torrent sequence adapters. These fragments are clonaly amplified on the Ion Sphere particles by emulsion PCR. The Ion Sphere particles with the amplified template are then applied to the `ion `torrent chip and the chip is placed in the ion PGM. The sequence run is set up on the PGM . Sequence results are provided in standard format. Downstream data analysis can then be performed.

11. Library preparation Genomic DNA Fragment shear reagents
Adapter compatable DNA
Ligate Adapters and nick-repair A P1
Adapters
Size-select unamplified library
(Preselected 200 or 100 base-reads)
1 Prepare good quality starting DNA
2 Enzymatically Fragment DNA with Ion Shear™ Plus Reagents and purify
3 Ligate Adapters and nick repair and purify the ligated DNA ( purify with
Agencourt® AMPure® magnetic beads)
4. Size-select the unamplified library with the E-Gel
5 determine if library amplification is required In general, 1 μg input libraries do not require amplification, and libraries from <100 ng inputs  do require amplification. A P1
Amplify the library

12. Size select fragments
Library size 200 base read- target peak size 330bp
1 Assess the size distribution of the library using bioanalyser
2. Determine the Template Dilution Factor that gives a concentration of ~26 pM using the bioanalyser or qPCR
The libraries are ready for the downstream Template Preparation procedure for clonal amplification  on Ion Sphere™ 
200 base read library gel
Assess the size distribution of the library using a bioanalyser

13. Check fragment size and DNA concentration
Critical Step to determine dilution factor for template preparation and assess the size distribution of library
Agilent Bioanalyzer with high sensitivity DNA Kit
Want dilution factor that give a concentration of approx 26pM
Want fragments of approx 330bp so peak at this and within small range of this
From[bp]To [bp] % ofTotal Average Size[bp] Size distribution inCV[ %]Conc.[pg/μl] Molarity[pmol/l]
, ,270.1

14. Template preparation
Libraries now ready for the downstream Template Preparation using the Ion One touch machine:
Adds fragments to ‘beads’ (ISPs)
in proportions that yield < one
fragment per bead.
Ion One Touch
DNA is fragmented, and each discrete DNA construct is amplified on a single bead—a process termed “clonal amplification.
Performs Emulsion PCR to
amplify the fragments on beads
aim- one fragment/bead – clonal amplification

15. Clonal Amplification

16. Clonal Amplification

17. J L
biotinilated
Excluded
(no target for
magnetic bead)

18. Load Chip with enriched particles
Wells designed to accommodate single bead only
Require sufficient amplified ‘loaded’ beads to occupy majority
of wells (confirmed by Quibt machine)
If earlier dilutions correct, should have few’mixed’ templates

19. Ion Torrent PGM Prep for every 2 runs: Clean (wash) PGM machine
Initialize machine and prepare solutions
For each sequencing run:
Anneal sequencing primer
Perform polymerase binding
Load the ion Chip
Sequencing run

20. The PGM Sequencing run
As mentioned earlier, individual dNTPs are passed over the chip one at a time, with each pass followed by a wash – An individual ‘pass + wash’ is called “a flow”
Four ‘flows’ are described as a ‘cycle’ – but every cycle is not identical – e.g. its not simply ATGC, ATGC, ATGC over and over again.
Instead, it repeats a specific set of 8 cycles (i.e. 32 flows) with each base represented 8 times – apparently this reduces systemic errors, but they do not explain how (?)

21. Flows and Cycles

22. Flows and Cycles

23. Flows and Cycles

24. Flows and Cycles

25. Ionograms

26. Data Analysis Export sequence data in suitable format to ‘end-user’
computer for other required analysis

27. Workflow – Realistic Times
Step Parallel numbers Time
DNA preps + checks 6 – 12 ½ day
Fragment library hr
Size selection + checks hr
Template prep hr
Enrichment ½ hr
PGM – Sequencing run ½ day*
*Doubled if 2 chips needed to provide sufficient coverage,

28. Costs Items cost/ bact genome (£) 200bp sequencing kit (£700) 70
200 bp OneTouch Systems kit (£700) 70
BioAnalyer DNA high sensitivity kit
[ and/or Q-PCR kit (£700)]
E-gels
Sequencing chip 314 (£70) x
[ or sequencing chip 316 x 1 @ > £200]
Overall – consumable costs minimum £300 – £400 per 314 chip run

29. Sequencing chips
Ion Torrent released semiconductor sequencing and the Personal Genome Machine™ sequencer at the end of 2010.sequencing.

30. Sequencing is just the start....
Analysis is also a bottleneck....

31. Comparison to run summary from IonTorrent literature
My run
Manufacturer’s run report
491, %
536, %
9, %
481, %
481,477
85, %
< 1%
61, %
335, %
1,262,519

32. Ion Torrent de novo analysis of Mycobacterium chelonae
Mycobacterium abscessus is an emerging pathogen
Part of the Mycobacterium chelonae clade in the fast growing mycobacteria
Both Mycobacterium chelonae and Mycobacterium abscessus can colonize cystic fibrosis patient’s lungs
When CF patient requires a lung transplant (late 20s early 30s) these mycobacteria cause problems
Mycobacterium chelonae can usually be treated
Mycobacterium abscessus usually cannot
Species are difficult to identify/distinguish

33. Ion Torrent de novo analysis of Mycobacterium chelonae

34. Ion Torrent de novo analysis of Mycobacterium chelonae

36. de novo assembly CLC Workbench Geneious Very fast, very expensive MIRA
Velvet
Very fast, very expensive
Very user friendly, cheap
Very thorough, free
User unfriendly, very high memory needs, free
Part of nice Virtual Box
PAGIT

37. M. chelonae vs M. abscessus with r2cat

38. M. chelonae vs M. abscessus with r2cat

39. Alignment of de novo contigs

40. Alignment using Abacas viewed in ACT

41. Gap is in alignment not in assembly

42. but still need another run with another ngs chemistry
Ion Torrent run gave x coverage
possibly too much
Different assemblers give slightly different contigs
I like MIRA for assembly
has active development and active user group
but using more than one assembler looks like a good strategy
Need a scaffolder to assemble contigs
GAP4 recommended but I’m still getting to grips
Abacas gives detailed alignment of contigs onto reference using ACT
r2cat gives easy to use mapping onto reference genome
but looks like have gaps
Ion Torrent library preparation is a lot of work

43. Minion Next generation sequencing?

44. Nanopore sensing

45. Label free strand sequencing

46. Comparison of ngs systems
Glenn (2013)

47. Minion Advantages Disadvantages Low capital cost (zero) High cost/Mb
Sample preparation No protocol yet
Long reads Not available!
Hairpin to read both strands
Same error rate along read 4% error rate (?)
Real time analysis Error not random
Multiplex samples

50. MAP Application form

51. Acknowledgements Alan Ward to Wendy Smith for Ion Torrent data
Wendy Smith to:
Dr. Jen Hallinan
Prof. Anil Wipat
Dr. Graham Colin
Prof. Colin Harwood
Dr. John Perry
Prof. Kate Gould