DNA sequencing: what's driving their improvements

How and Why are Costs of DNA Sequencing Falling?

What are the Implications of these falling costs? 6th Session of MT5009

A/Prof Jeffrey FunkDivision of Engineering

and Technology ManagementNational University of Singapore

For information on other technologies, see http://www.slideshare.net/Funk98/presentations

Objectives

What are the important dimensions of performance for DNA sequencers and higher-level systems?

What are the rates of improvement? What drives these rapid rates of

improvement? Will these improvements continue? What kinds of new higher-level systems

will likely emerge from the improvements in DNA sequencers?

What does this tell us about the future?

Session Technology

1 Objectives and overview of course

2 Two types of improvements: 1) Creating materials that better exploit physical phenomena; 2) Geometrical scaling

4 Semiconductors, ICs, electronic systems

5 MEMS and Bio-electronic ICs

6 Nanotechnology and DNA sequencing

7 Superconductivity and solar cells

8 Lighting and Displays

9 Human-computer interfaces (also roll-to roll printing)

10 Telecommunications and Internet

11 3D printing and energy storage

This is Part of the Sixth Session of MT5009

Creating materials (and their associated processes) that better exploit physical phenomenon

Geometrical scaling◦ Increases in scale◦ Reductions in scale

Some technologies directly experience improvements while others indirectly experience them through improvements in “components”

As Noted in Previous Session, Two main mechanisms for improvements

A summary of these ideas can be found in 1) What Drives Exponential Improvements? California Management Review, Spring 2013 2) Technology Change and the Rise New Industries, Stanford University Press, 2013

Creating materials (and their associated processes) that better exploit physical phenomenon◦ Created materials that enable new techniques of DNA

sequencing Geometrical scaling

◦ Reductions in scale: smaller feature sizes for each technique (but many new techniques)

◦ Increases in scale: larger wash plates and production equipment

Some technologies directly experience improvements while others indirectly experience them through improvements in “components” ◦ Better lasers and sensors were important for some of the

techniques (e.g., pyrosequencing and Single-molecule real-time sequencing)

Both are Relevant to DNA Sequencing

Identify the sequence and identity of 3 billion base pair nucleotides in DNA strand

Nucleotides encode the genetic instructions for organisms

Four types of nucleotides in a DNA strand◦ Adenine◦ Thymine◦ Cytosine◦ Guanine

The Challenge

http://www.genome.gov/sequencingcosts/

http://www.genome.gov/sequencingcosts/

Read lengths Accuracies Speeds Improvements in these variables also lead to reductions in cost of sequencing

Capability to analyze and use gathered data◦need better computers◦need more storage

Improvements have also occurred in…

Improvements in DNA sequencers

Nature 2011, 470: 198-203, Elaine Mardis

Why do Costs Fall? New Methods Continue to Emerge

◦ Pyrosequencing (454 Life Sciences/Roche and Illumina)

◦ Single-molecule real-time sequencing (Pacific Bio)

◦ Semiconductor arrays (Ion Torrent)◦ Nanopores (Oxford Nanopore Technologies)◦ Methods of data compression

Synthesizing DNA Who Cares? What are the Implications? Conclusions

Outline

New methods of sequencing◦ Maxam-Gilbert Sequencing: relies on cleaving of

nucleotides by chemical methods ◦ Chain Termination methods (sometimes called

Sanger method): bases are illuminated with UV light, read with X-rays

◦ Dye-termination: reading sequences with fluorescent dyes where each nucleotide emits light in different wavelengths (this technology caused acceleration)

Improved lasers and cameras to read fluorescent dyes

More parallel processing Smaller feature sizes, reductions in scale

Why do Costs Fall?

http://www.dnasequencing.org/history-of-dna

Source: Nature Biotechnology 30(11), 1023-1026, November 2012

But many different approaches are being investigated

This can be understood by reading highly cited papers such as◦ “Genome sequencing in micro-fabricated high-density

pico-liter reactors” (Margulies, 2005) and◦ “Toward nano-scale genome sequencing” (Ryan et al,

2007). Quote from Ryan et al: “The ability to construct nano-scale

structures and perform measurements using novel nano-scale effects has provided new opportunities to identify nucleotides directly using physical, and not chemical, methods.”

In fact, just the titles of these papers are fairly suggestive. In all of these decreasing scale examples, totally new forms of equipment, processes and factories were required.

Newer Approaches have smaller scale






Outline

1) separate DNA into smaller strands 2) make copies of strands (i.e., amplification) with

emulsion beads in plastic containers ◦ do this with small containers on a large wash plate so that many

copies are made in parallel◦ smaller containers and larger wash plates lead to more parallel

and faster processing 3) identify DNA nucleotides utilizing lasers and cameras

◦ Nucleotides emit light in the presence of an enzyme, ADT (Adenosine Triphosphate)

◦ falling costs of lasers and cameras reduce costs 4) Analyze data with computers

Pyrosequencing by 454 Life Sciences

One source: http://www.454.com/downloads/news-events/how-genome-sequencing-is-done_FINAL.pdf

Make copies to improve accuracy through redundancy

454 PicoTiterPlate from LifeSciences◦ contains 1.6 million hexagonal wells◦ each holds 75 pico-liters (10-12 liters, <100 micron diameter)

These wells can be made much smaller◦ dimensions on integrated circuits (ICs) are on the order of

20 nano-meters◦ Is it possible to reduce feature sizes by 1000 times or

volumes by 109

Make Copies of Strands (i.e., amplification in Step 2)

Fluorescent Dyes, Lasers, and Cameras

(Step 3)

As bases move across wash plate during sequencing run, a nucleotide (molecules that make up DNA) generates light signal, which is recorded by camera

Signal strength is proportional to number of nucleotide incorporated onto the DNA strands






Outline

Eliminate amplification and wash steps with zero wave guides (Pacific BioSciences)

http://www.youtube.com/watch?v=v8p4ph2MAvI from 1:50 to 3:50

http://www.youtube.com/watch?v=v8p4ph2MAvI

Uses Zero Mode Wave Guides They are

◦Very small: zepto-liters (10-21 liters, 50 nanometers in diameter)

◦fabricated in a 100nm metal film on a silicon dioxide substrate

◦enough room for 600,000 molecules of liquid water at room temperature

◦How much smaller can they be made?

Pacific Bio-Sciences






Outline

Uses semiconductor chips to sequence DNA by detecting PH differences between A, G, C, and T◦ Thus, no lasers, cameras, or amplification are used

A microwell containing template DNA strand is filled with single species of deoxyribonucleotide triphosphate (dNTP)◦ Beneath layer of microwells is ion sensitive layer, below

which is ISFET ion sensor.

◦ All layers are contained in CMOS semiconductor chip

◦ If the introduced dNTP is complementary to leading template nucleotide, it is incorporated into growing strand

◦ This causes release of a hydrogen ion that triggers ISFET ion sensor, indicating a reaction has occurred

Ion Torrent

http://www.nature.com/news/2010/101214/full/news.2010.674.html http://en.wikipedia.org/wiki/Ion_semiconductor_sequencinghttp://www.lifescientist.com.au/article/394936/feature_sequencing_3_0/?pp=2

Done in Massively Parallel For each well

Matches cause ion to be released

Multiple matches cause multiple ions to be released

No matches no ions are released

While first sequencers used older (i.e., large feature sizes) semiconductor technology, newer ones use smaller feature sizes and thus are faster than older ones http://www.youtube.com/watch?v=JHzkYDyMzOg&feature=relmfu (2:30-

4:15)

For example, first sequencer (314) had 1.2 million wells while most recent one (Proton II) has 660 million wells◦ How much smaller can these wells be made? ◦ Since 256GB memory chips (1 byte = 8 bits) exist, can ion torrent

be able to provide 256 x 8 billion wells or about 2 trillion wells in next few years?

◦ After that improvements may slow as ion torrent's improvements depend on reductions in feature sizes of semiconductor technology

Ion Torrent

http://www.nature.com/news/2010/101214/full/news.2010.674.html http://en.wikipedia.org/wiki/Ion_semiconductor_sequencinghttp://www.lifescientist.com.au/article/394936/feature_sequencing_3_0/?pp=2

Source: Ion Torrent Video






Outline

Squeeze DNA through a nano-scopic pore (about 1.4 nm) in a semiconductor and read the distinctive change each letter in the sequence makes in the amount of current flowing through the pore

NanoPores

DNA moves through a nanopore at remarkably high velocities and thus only a small number of ions (as few as ~100) are available in the nanopore to correctly identify nucleotides◦ so the small changes in the ionic current due to the

presence of different nucleotides are overwhelmed by thermodynamic fluctuations

Challenge is to reduce the translocation velocity so that the ions can be correctly identified

Reductions in Translocation Velocity

http://www.youtube.com/watch?v=wvclP3GySUY

http://www.nature.com/nnano/journal/v6/n10/fig_tab/nnano.2011.129_F1.html

nt=

nucl

eoti

des

Reductions in Translocation Velocity over Time

2000 nanopore system (900 USD) that can read DNA at a rate of hundreds of kilobases per second

8000 nanopore system by next year (2013) that can read more than 1M bases per second

With about 3 billion bases per human genome and 20 sequencing machines, it takes about 15 minutes to sequence human genome

Expected Sequencing Time of 15 minutes with Oxford Nanopores

http://www.nature.com/news/nanopore-genome-sequencer-makes-its-debut-1.10051

The reduced velocities (and improved sensitivities) achieved by◦ combination of site-specific mutagenesis and one of

the following: the incorporation of DNA processing enzymes into the nanopore, chemical labeling of the nucleotides or the covalent attachment of an aminocyclodextrin adapter for α-haemolysin

◦ optimization of solution conditions (temperature, viscosity, pH), chemical functionalization, surface-charge engineering, varying the thickness and composition of the membranes, and the use of smaller diameter nanopores (thereby enhancing polymer–pore interactions) for solid sate

Further Reductions in Translocation Velocity

http://www.nature.com/nnano/journal/v6/n10/fig_tab/nnano.2011.129_F1.html

Personal Sequencing, Garage Biology

Sequencing can be done in your home, office, or in field

Sequence your own DNA multiple times in your life

Sequence the DNA from a bucket of ocean water, sewage, or

handful of dirt

Find proteins to manufacture other things

Combined with 3D printers, PCs, and the Internet, there is no limit to what we can do as individuals $900 from Oxford Nanopore






Outline

Many believe this will be the bottleneck in genome sequencing

Partial solution: because there are redundancies in the data, better algorithms can speed up the sequencing

Amount of Data is Exploding

Source: Nature 498 pp. 255-260, 13 June 2013

a) File sizes of the uncompressed, compressed with links and edits, and unique sequence data sets with default parameters. (b) Run times of BLAST, compressive BLAST and the coarse search step of compressive BLAST on the unique data ('coarse only'). Error bars, s.d. of five runs. Reported runtimes were on a set of 10,000 simulated queries. For queries that generate very few hits, the coarse search time provides a lower bound on search time. (c) Run times of BLAT, compressive BLAT and the coarse search step on the unique data ('coarse only') for 10,000 queries.

Cloud Computing

For storage and processing How to encourage sharing of data? How to protect privacy? Who will be the leading providers and users of these services?

How will this impact on the overall industry of health care?◦Might this globalize health care?






Outline

We can synthesize new forms of DNA Make new drugs, crops, or materials Test them Then synthesize/design newer forms of DNA

Keep iterating and making better drugs, crops, and materials

Synthesizing DNA

The cost of synthesizing DNA is also dropping

http://singularityhub.com/2012/09/17/new-software-makes-synthesizing-dna-as-easy-as-using-an-ipad/

http://www.synthesis.cc/cgi-bin/mt/mt-search.cgi?blog_id=1&tag=Carlson%20Curves&limit=20

About 5 years behind sequencing






Outline

Most drugs are naturally occurring substances But improvements in our knowledge of humans

and other organisms and reductions in cost of sequencing and synthesizing DNA increase possibility of synthesizing drugs◦ Begins with DNA "target”: naturally existing cellular or

molecular structure involved in pathology of interest ◦ A common target is proteins whose function has now

become clear as a result of basic scientific research ◦ Sequence the protein’s DNA and then synthesize a drug

that acts on this protein (also based on scientific research)

Drug Discovery (1)

Gary Pisano, Science Business: The promise, the reality, and the future of biotech, Chapters 2 and 3

If we can reduce the cost of drug development, we can target smaller groups of people with drugs

How about synthesizing drugs for individuals? How about understanding which diseases a human

might be susceptible by sequencing their DNA? Even if we cannot synthesize drugs for individuals,

maybe we can better assign drugs to individuals by better understanding which humans are susceptible to known side effects◦ Most drugs have side effects

◦ DNA can tell us who might be susceptible to the side effects

Drug Discovery (2)

Gary Pisano, Science Business: The promise, the reality, and the future of biotech, Chapters 2 and 3

Gleevec treats myeloid leukemia◦ Blocks activity of protein BCR-ABL; it comes from abnormal

gene created by a merge of chromosomes 9 and 22 Crizitnonib teats lung cancer

◦ mutated version of gene called ALK, encodes protein that instructs lung cells to divide uncontrollably

Vemurafenib treats melanoma◦ Attacks protein that is generated by mutated version of a

gene called BRAF Problems

◦ Many cancers driven by more than one mutation and genes involved in repair are often involved with mutations

◦ $100,000 for 4 doses of one drug Nevertheless, DNA sequencing is helping scientists

identify common genes for cancer

Examples

Source: Getting Close and Personal, Economist, January 4, 2014

Green Machines for Better Crops?

Better sensors (cameras, infrared, fluorescence, lasers) and mechanical controls enable complete control and measurement over crop growth

DNA sequencing and DNA synthesizing enable characterization and replication of high performing crops

Other biological materials? Cellulosic ethanol Algae

http://www.aber.ac.uk/en/media/departmental/ibers/facilities/phenomicscentre/BBC-FOCUS-NPPC-Feature.pdf

Source: https://www.soils.org/publications/cs/articles/46/2/528

Improvements in U.S. Corn Yields through New Seeds

Improvements in Yield for other Crops

U.S. Department of Agriculture and Michael Bomford, Crop Yield Projections

According to Science Magazine, Scale-up will not enable economic feasibility

It’s not just about ◦ making bio-fuels from the non-food part of the

plant or◦ scaling up the production in order to reduce cost

It’s also about Developing Better Organisms◦ Better cellulose that produces more ethanol per

weight, while still enabling the plant to produce lots of food

◦ Better algae that consumes more carbon dioxide and generates more energy per weight or area

Better Bio-fuels

http://www.theguardian.com/science/2012/jan/14/synthetic-biology-spider-goat-genetics

Spider silk is very strong But difficult to harvest spider silk, partly

because it is hard to raise spiders (they eat each other)

Scientists introduced the gene for spider silk into goats so spider silk would be produced in their milk

Now spider silk is produced in the goat’s milk and scientists are trying to improve the results

It is expected that many other natural substances can be manufactured in this way

Synthesize Better Materials

http://www.theguardian.com/science/2012/jan/14/synthetic-biology-spider-goat-genetics

Enzymes, plastics, textiles, dyes Many of these are now made from fossil

fuels but were once made form natural substances

Can we return to biological feedstocks?◦ Modify yeast so that sugar can be turned into

useful compounds such as malaria drugs and biofuels

◦ Bring a switch from fossil fuels to biological feedstocks such as sugar, starch, and cellulose

Other Types of Materials

Registry of Standard Biological Parts◦ More than 10,000 parts

Can build complex systems from these parts

Genetically Engineered Machine Competition◦ Students compete to build complex systems◦ One group built a biological light detector with a

resolution of 100 million pixels per square inch Will biological systems ever compete with

electronic systems?

Taking all of this One Step Further:Building Complex Systems

DNA synthesizing equipment can be used to make (and replicate) DNA

One challenge is how to insert DNA into a cell, so that the cell can then replicate itself◦ Each cell contains DNA needed for a specific organism◦ Each cell may even contain the DNA for features that no

longer exist and the features can be turned back on

First done by Craig Venter’s team in May 2010◦ His team synthesized an entire bacterial genome

and “took over” a cell by inserting the DNA into the cell

Can this be done for more complex life forms?

How About “Creating” Life?

Source: Michio Kaku, Physics of the Future: How Science Will Shape

Human Destiny and Our Daily Lives by the Year 2100 (2011)

More Complex Organisms Require More Base Pairs and thus more years for their Synthesizing

The cost of sequencing and synthesizing DNA continues to fall

A major reason for the cost reductions is the benefits from reductions in scale◦Similar to those in ICs, bio-electronic ICs,

and MEMS◦A powerful way to reduce costs

Further reductions in scale and thus further cost reductions appear possible

Conclusions (1)

Low cost and small DNA sequencers and synthesizers will change drug discovery, health care, and science ◦ How will we do drug discovery in the future?

What kind of analyses can help us understand how these trends will change drug discovery and health care?

What kinds of opportunities will emerge for firms as vast amounts of data become available for analysis?

Conclusions (2)

Appendix

Single cell genomics ◦select the embryos created by IVF (in

vitro fertilization) that have best chance of developing into a healthy baby

Metagenomic medicine◦Sequencing many different microbes en

masse and then teasing out individual genomes to diagnose which ones are helping or harming human health

Specific Changes for in vitro fertilization

Nature 494, 21 February 2013, pp. 290-291