Silicon meets Biotech: Building Better Computers by Incorporating Biological

Parts !

Luis!Ceze!!!

University!of!Washington!

100,000x%%in%8%years!%

The image cannot be displayed. Your computer may not have enough memory to open the image, or the image may have been corrupted. Restart your computer, and then open the file again. If the red x still appears, you may have to delete the image and then insert it again.

Computer%Industry%

Biotech%Industry%

Helped&a&lot!& Time&for&payback!&

Biology 3D Density Self-assembly Efficiency Sensitivity Can sense things silicon can’t&

Silicon%Speed Engineerability Integration with Infrastructure

Storage

5&

11011101 11011101 AGCTATCAG& AGCTATCAG&

write&path& read&path&

107&poten>al&improvement&over&tape&

Addressing&&Redundancy&System&overheads&&&

And&readers&never&become&obsolete!&3F5& 5& 10F30& 1000+%100&

Life>me&&(years)&

0.05&Archive&Cost&($/GB)& 200.00&

Storage

Photo:&Tara&Brown&/&UW&10TB%

Encoding digital data in DNA The&“direct”&way:&convert&long&bit&string&to&base&4&

101000111001000111100111110001011001010010111101…&

&2&2&0&3&2&1&0&1&3&2&1&3&3&0&1&1&2&1&1&0&2&3&3&1&

&G&G&A&T&G&C&A&C&T&G&C&T&T&A&C&C&G&C&C&A&G&T&T&C&

A&0&

C&1&

G&2&

T&3&

Direct mapping isn’t good enough

&1.  Repeated&leWers%are&bad:&&&&

Use&base&3&and&“rotate”&mapping.&

G& G& A& T& G& C& A&

P[AWach]&=&99%&

99%& 98%& 97%& 96.1%& 95.1%& 94.2%&

100&nts&

36.6%&

200&nts&

13.4%&

…& …&

2.  Synthe>c&DNA&sequences&have&limited&length:&&

The&direct&mapping&approach&isn’t&appropriate&for&two&reasons:&

Breaking up data into chunks

~%10%bytes%per%DNA%strand.%%%Many%strands%per%file.%

…%

1%of%N%

2%of%N%

3%of%N%

&A&A&A&A&

&A&A&A&C&

&A&A&A&G&Addresses&within&the&file&

&C&A&T&C&C&

&C&A&T&C&C&

&C&A&T&C&C&

&A&T&G&T&T&

&A&T&G&T&T&

&A&T&G&T&T&File&iden>fiers&(“primers”)&

&G&C&A&C&&G&G&A&T&

&T&G&C&T& &T&A&C&C&

&G&C&C&A& &G&T&T&C&

Payload&chunks&

Errors in writing/reading DNA

G& G& A& T& A& G& C&

G& G& A& T& G& C& A&

G& G& A& T& G& A&

G& G& A& T& C& C& A&

InserIons%

DeleIons%

SubsItuIons%

A&

Aggregate&error&rates&~1%&

Adding redundant information

Reserved&address&space&to&indicate&redundancy&data&

&C&A&T&C&C&

&G&C&A&C&&G&G&A&T&

&T&G&C&T& &T&A&C&C&

&G&C&C&A& &G&T&T&C&

&A&A&A&A&

&A&A&A&C&

&A&A&A&G&

&C&A&T&C&C&

&C&A&T&C&C&

&A&T&G&T&T&

&A&T&G&T&T&

&A&T&G&T&T&

Addresses&within&the&value&

Key&iden>fiers&(“primers”)&

Payload&chunks&

&C&A&T&C&C&

&T&C&G&C& &T&A&C&G&

&T&G&C&A& &A&T&T&C&

&C&A&A&C&

&CA&A&G&

&C&A&T&C&C&&A&T&G&T&T&

&A&T&G&T&T& &C&A&T&C&C&

Redundant&DNA&strands.&Many&possibili>es:&Reed&Solomon,&LDPC,&etc&

Random access?

?&

&C&A&T&C&C&

Random access!

PCR&

&Selec>vely&amplify&strands&based&on&their&primer&

Sample&Strands&with&3&different&primers&

Almost&all&strands&have&desired&primer&

&T&A&T&C&T&

&G&C&A&C&&G&G&A&T&

&T&G&C&T& &T&A&C&C&

&G&C&C&A& &G&T&T&C&

&A&C&T&A&

&G&A&T&C&

&A&G&C&G&

&G&A&T&A&C&

&A&T&G&T&T&

&C&C&A&C&T&

&A&T&G&T&T&&T&A&C&A&A& &A&T&A&G&A&

File&iden>fiers&(“primers”)&

Experiments catcatgg&

catcatgc%

Throughput&10s&MB/week&

200MB as of July’16. (1.5 B nucleotides)

A DNA-based archival storage system

101001101…&

[ASPLOS’16]&

Molecular Information Systems Lab

Computer architects, coding theorists, biochemist, molecular biologists, fluidics, algorithms, …

DNA-based Archival

Tape

HDD

Flash

HalfFlife&of&DNA&in&bone:&~521&years.&

Durable storage by nature

Proteins&are&amazing&machines!&

An>body&Enzymes&(carry&out&most&reac>ons)&Messenger&(e.g.,&hormones)&Structure&Transport/storage&(e.g.,&oxygen,&ions)&

1%ATP%molecule%%=%%%%~%10M21%%J%

Processing

A&“proteinistor”&would&be&10,000x&more&energy&efficient&than&a&CMOS&transistor&

ATP powered “supercomputer”?

Parallel computation with molecular-motor-propelledagents in nanofabricated networksDan V. Nicolau Jr.a,b,1, Mercy Lardc,1, Till Kortend,e,1, Falco C. M. J. M. van Delftf,2, Malin Perssong, Elina Bengtssong,Alf Månssong, Stefan Diezd,e, Heiner Linkec,3, and Dan V. Nicolauh,i,3

aDepartment of Integrative Biology, University of California, Berkeley, CA 94720-3140; bMolecular Sense, Ltd., Wallasey CH44 1AJ, United Kingdom;cNanoLund and Solid State Physics, Lund University, S-22100 Lund, Sweden; dCenter for Molecular Bioengineering (B CUBE) and Center for AdvancingElectronics Dresden (cfaed), Technische Universität Dresden, 01069 Dresden, Germany; eMax Planck Institute of Molecular Cell Biology and Genetics, 01307Dresden, Germany; fPhilips Research (MiPlaza) and Philips Innovation Services, 5656 AE, Eindhoven, The Netherlands; gDepartment of Chemistry andBiomedical Sciences, Linnaeus University, SE-39182 Kalmar, Sweden; hDepartment of Electrical Engineering & Electronics, University of Liverpool, LiverpoolL69 3GJ, United Kingdom; and iDepartment of Bioengineering, McGill University, Montreal, QC, Canada H3A 0C3

Edited by Hillel Kugler, Microsoft Research, Cambridge, United Kingdom, and accepted by the Editorial Board January 18, 2016 (received for review June5, 2015)

The combinatorial nature of many important mathematical problems,including nondeterministic-polynomial-time (NP)-complete problems,places a severe limitation on the problem size that can be solved withconventional, sequentially operating electronic computers. There havebeen significant efforts in conceiving parallel-computation approachesin the past, for example: DNA computation, quantum computation,and microfluidics-based computation. However, these approacheshave not proven, so far, to be scalable and practical from a fabricationand operational perspective. Here, we report the foundations of analternative parallel-computation system in which a given combinato-rial problem is encoded into a graphical, modular network that isembedded in a nanofabricated planar device. Exploring the network ina parallel fashion using a large number of independent, molecular-motor-propelled agents then solves the mathematical problem. Thisapproach uses orders of magnitude less energy than conventionalcomputers, thus addressing issues related to power consumptionand heat dissipation. We provide a proof-of-concept demonstrationof such a device by solving, in a parallel fashion, the small instance{2, 5, 9} of the subset sum problem, which is a benchmark NP-complete problem. Finally, we discuss the technical advances neces-sary to make our system scalable with presently available technology.

parallel computing | molecular motors | NP complete | biocomputation |nanotechnology

Many combinatorial problems of practical importance, suchas the design and verification of circuits (1), the folding (2)

and design (3) of proteins, and optimal network routing (4),require that a large number of possible candidate solutions areexplored in a brute-force manner to discover the actual solution.Because the time required for solving these problems grows ex-ponentially with their size, they are intractable for conventionalelectronic computers, which operate sequentially, leading to im-practical computing times even for medium-sized problems.Solving such problems therefore requires efficient parallel-com-putation approaches (5). However, the approaches proposed so farsuffer from drawbacks that have prevented their implementation.For example, DNA computation, which generates mathematicalsolutions by recombining DNA strands (6, 7), or DNA static (8) ordynamic (9) nanostructures, is limited by the need for impracticallylarge amounts of DNA (10–13). Quantum computation is limited inscale by decoherence and by the small number of qubits that canbe integrated (14). Microfluidics-based parallel computation (15) isdifficult to scale up in practice due to rapidly diverging physical sizeand complexity of the computation devices with the size of theproblem, as well as the need for impractically large external pressure.Here, we propose a parallel-computation approach, which is

based on encoding combinatorial problems into the geometry of aphysical network of lithographically defined channels, followed byexploration of the network in a parallel fashion using a largenumber of independent agents, with very high energy efficiency.

To demonstrate operational functionality, we applied it to a smallinstance of a benchmark classical nondeterministic-polynomial-timecomplete (NP-complete) problem (16), the subset sum problem (SSP)(Fig. 1). This problem asks whether, given a set S = {s1, s2, ..., sN}of N integers, there exists a subset of S whose elements sum to atarget sum, T. More formally, the question is whether there is asolution

PNi=1wisi where wi ! {0, 1}, for any given T from 0 toPN

i=1si. To find all possible subset sums by exploring all possiblesubsets requires the testing of 2N different combinations, which––even for modest values ofN––is impractical on electronic computersbecause of exponentially increasing time requirements (SI Appendix,section S1). Although more sophisticated algorithms exist (17–19),none of these avoids the exponentially growing exploration time, aproperty that is harnessed in some cryptography systems togenerate encoded messages (20).

Significance

Electronic computers are extremely powerful at performing a highnumber of operations at very high speeds, sequentially. However,they struggle with combinatorial tasks that can be solved faster ifmany operations are performed in parallel. Here, we presentproof-of-concept of a parallel computer by solving the specificinstance {2, 5, 9} of a classical nondeterministic-polynomial-timecomplete (“NP-complete”) problem, the subset sum problem. Thecomputer consists of a specifically designed, nanostructured net-work explored by a large number of molecular-motor-driven,protein filaments. This system is highly energy efficient, thusavoiding the heating issues limiting electronic computers. Wediscuss the technical advances necessary to solve larger combi-natorial problems than existing computation devices, potentiallyleading to a new way to tackle difficult mathematical problems.

Author contributions: Dan V. Nicolau Jr. and Dan V. Nicolau conceived the calculationmethod and designed the overall network; F.C.M.J.M.v.D designed the junctions; M.L.,T.K., F.C.M.J.M.v.D, A.M., S.D., and H.L., designed the device layouts; M.L. and F.C.M.J.M.v.Dfabricated the devices; M.L., T.K., M.P., and E.B. ran motility experiments and ana-lyzed motility data; Dan V. Nicolau Jr., T.K., and A.M. carried out numerical simula-tions; Dan V. Nicolau initiated the project; Dan V. Nicolau and H.L. coordinated theproject; and Dan V. Nicolau Jr., M.L., T.K., F.C.M.J.M.v.D., M.P., E.B., A.M., S.D., H.L.,and Dan V. Nicolau contributed to planning the work, to data interpretation, and towriting the manuscript.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission. H.K. is a guest editor invited by the EditorialBoard.

Freely available online through the PNAS open access option.1Dan V. Nicolau Jr., M.L., and T.K. contributed equally to this work.2Present address: Molecular Sense, Ltd., Wallasey CH44 1AJ, United Kingdom.3To whom correspondence may be addressed. Email: heiner.linke@ftf.lth.se or dan.nicolau@mcgill.ca.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1510825113/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1510825113 PNAS | March 8, 2016 | vol. 113 | no. 10 | 2591–2596

COMPU

TERSC

IENCE

SBIOPH

YSICSAND

COMPU

TATIONALBIOLO

GY

Or an ATP battery?

10x%energy%density%of%liMion%baPeries%

Molecular-Level Self-Assembly and Reconfiguration

DNA&origami&

35.8 nm&

Proteins&

[Caltech,&Duke]&

Single-Molecule Sensing

22&

Gu&et&a

l.&Single&m

olecule&sensing&by&nanop

ores&and

&nano

pore&devices.&2009,&Analyst.&

C& G&T&G&C&G&A&G&G& C& C& T&

What about output?

CuPlePhone%

Silicon + Biotech has immense opportunities!

Inevitable to go to the molecular/atomic level.

Biology has “invented” many useful parts, not just ideas.

Biotech is quickly developing many useful tools/methods.