This document discusses nanopore sequencing technology from Oxford Nanopore Technologies. It provides details on their MinION and PromethION sequencing devices, including the design of the MinION flow cell and basecalling process. It also describes the MinION Access Program (MAP) and MinION Analysis and Reference Consortium (MARC) for evaluating and improving the nanopore sequencing platform. While showing promise, the document notes some areas still needing improvement for the technology to be fully ready for production, including flow cell quality and throughput.

1. Nanopore Sequencing
Hans Jansen
ZF-screens B.V.

2. ZF-screens B.V.
Common carp (Cyprinus carpio)
High throughput screening model
Genome and transcriptomes
European and Japanese eel (Anguilla anguilla and Anguilla japonica)
Completing the life cycle in aquaculture
Genome and transcriptomes
King cobra (Ophiophagus hannah)
Evolution and toxins
Genome and transcriptomes
But the quality of these genomes could be improved.
Dutch SME located in the Bioscience Park in Leiden, the Netherlands.
Compound screens, Fish fertility. Sequencing.

3. • Started to work on nanopore sensing in 2005
• Investments to date 251 million GBP (355 M€)
• Valued at 1.2 B€.
• Broad IP portfolio on nanopore sensing.
• Products: MinION and PromethION systems
Oxford Nanopore Technologies

4. But MAP is much more. It is about being a community and a playground to test new
applications and analysis tools.
Visible as a web portal with information from ONT and social media like system with
blog possibilities, comment, likes, and a forum to ask advice.
MinION Access Program

5. Inlet port
Electrode
Array of 2048 wells with nanopores embedded in
polymer membrane.
Underneath is an Application Specific Integrated Circuit
(ASIC) which contains the 512 signal amplifiers (SA).
MinION Flow Cell design
Waste reservoir
SA
well
After 24 hrs a well is exhausted and the
SA switches to a new well (remux).

6. Inlet port
Electrode
Array of 2048 wells with nanopores embedded in
polymer membrane.
Underneath is an Application Specific Integrated Circuit
(ASIC) which contains the 512 signal amplifiers (SA).
MinION Flow Cell design
Waste reservoir
SA
well
After 24 hrs a well is exhausted and the
SA switches to a new well (remux).

7. Stills taken from: https://www.nanoporetech.com/news/movies#movie-24-nanopore-dna-sequencing
The ion flux is partially blocked by the translocating DNA strand.
Potential (voltage) is applied across the membrane and there is a ion flux through the pore.
The ion flux is measured by an application-specific integrated circuit (ASIC).
+
-
Strand sequencing
ATP
ADP

8. Tethering oligo
Motor protein
Hairpin
abasic nucleotides
• Tethers keep the DNA molecules on the membrane to increase concentration of fragments.
• Motor protein unwinds ds DNA and ratchets it though the pore at a controlled rate.
• Abasic nucleotides in the hairpin are a marker for the basecaller to separate template and
complement data.
Tethering oligo
Adapter
Template strand
Complement strand
Strand sequencing
Template Complement

9. Bulk Data (ionic curent, pA)
Events (with time domain)
Squiggle (events with time domain removed)
~6 nucleotides are blocking the current when a DNA strand is in the pore, but
nucleotides around this k-mer can influence the signal.
AGGCTC
AGGCTCACTCCCATAAGC
A
C
G
C
G
T
A C
G
T
A
C
G
T
A
Base calling
T
Reading in k-mer space

10. AGGCTC
AGGCTCACTCCCATAAGC
A
C
G
C
G
T
A C
G
T
A
C
G
T
A
Base calling
T
Reading in k-mer space
AGGCTCACTCCCATAAGC
AGGCTC
GGCTCA
GCTCAC
CTCACT
TCACTC
CACTCC
ACTCCC
CTCCCA
The A is measured 6 times on the template strand and
the corresponding T 6 times on the complement strand.
Event
Event
Event
Event
Within an event the different possibilities have an
probability value attached to them. Base calling is
choosing a path through all possibilities with the
highest probability.

11. Read length is limited by the non-nicked fragment length rather than the by the system itself.
Longest 2D reads are over 100 Kbp, 1D reads over 200 Kbp.
Read length distribution

12. The base caller should model the translocation process in and around the pore.
Most errors are a result of the basecaller not being accurate enough.
Error rate lies around ~10-13% for current chemistry (R7.3 and MAP006). The contribution of
deletions, insertions and substitutions is more or less equal.
Challenging are still:
Homopolymers Time domain of the events might be used to resolve this.
DNA modifications Modified bases need to be added to the model.
AT rich area’s Better nanopores and better modeling.
Base caller code will be released shortly to enable the community to develop different base
callers.
In MARC data will be generated to enable the incorporation of modified bases in the base
calling model.
The R9 nanopore is in the pipeline (improving on G/C bias and better S/N).
Errors

13. GC content
June 2014 February 2016
Abundance of k-mers in the reads is plotted against the abundance in the reference

14. MinION Analysis and Reference Consortium
The MinION Analysis and Reference Consortium (MARC) is a
collaboration of members of MAP.
• Evaluate the reproducibility of the MinION platform
• Investigate improvements to the library preparation and running
protocols.
• Phase 1: 5 labs running the same protocols on the same samples.
A first paper was published on F1000research.com in a channel dedicated to the analysis
of nanopore sequence data (Ip, L. C., et al. doi: 10.12688/f1000research.7201.1)
http://f1000research.com/channels/nanoporeanalysis
MARC intends to publish more papers in this channel.
This channel is open to all papers that fall within the scope of the channel.

15. Reproducibility: flow cell yield

16. Active pores during a run
remux 24 hrs remux 48 hrs

17. Reproducibility: active pores
Percentage of active pores decreases during the run.
Amount of active pores can vary significantly.

18. Reproducibility: read length
Read length is influenced by the shearing conditions, DNA quality, g-TUBE.

19. We joined MAP right from the start.
Our first MinION arrived in April 2014 and the first flow cells and kits in June.
Since then we run ~50 Flow Cells.
MAPpers competition
Topped the leader board on read length and yield so we now have three MinION's.
MinION Access Program and ZF-screens
And one of these
is on the way

20. 144000 channels producing 6.4 TB/day @ 500 bps/sec.
Basecalling is done locally.
Experimental design can be flexible. A loading port is connected to 750
channels (4 ports/Flow Cell). Flow Cells can be run independently from each
other.
PromethION

21. Conclusions
The Nanopore platform is very promising but its not completely ready for production.
The MinION is a platform with some unique characteristics.
• Very mobile.
• Cost structure is very different.
• Reads can be very long.
• Data can be accessed in real time.
Things that should be improved
• Flow cell quality.
• Base calling.
• Throughput.
What I would like to see happening
• Development of tools that work with the event data rather than the base called data.
• This might also allow for “streaming analysis”.

22. Ron Dirks (CEO of ZF-screens B.V.)
All members of the MARC consortium
Ewan Birney, EMBL-EBI
Camilla Ip, WTCHG Oxford
John Tyson, University of British Columbia
Justin O’Grady, UEA
Sara Goodwin, CSHL
Vadim Zalunin, EMBL-EBI
Miten Jain, UCSC
Matt Loose, Nottingham
Jared Simpson, OICR, Toronto
Acknowledgements
Rosemary Dokos
Oliver Hartwell
Christiaan Henkel (Assistant professor Leiden University)