Computing and Biomolecules Alvin R. Lebeck Duke University Partial - PowerPoint PPT Presentation

Computing and Biomolecules Alvin R. Lebeck Duke University

Partial Goals of Talk • Introduce you to potentially disruptive technology • Opportunities & Challenges • Challenge you to think “outside the box” • Maintain vs. break abstractions • Bridge the Engineering Gap • Back to the Future: understand entire stack from chemistry/physics up through applications (hipster architect?) • Be interdisciplinary! 2

Setting Context • Computing • Processing and storing information • Biomolecules • DNA, proteins, fluorescent molecules, etc. • Everyday use in the Life Sciences Why put these together? 1. How do we put these together? 2. • First some background on biomolecules 3

Biomolecules: Synthetic DNA • Single strand is sequence (T) Thymine Adenine (A) of nucleotides Cytosine (C) (G) Guanine • Well defined rules for base pair matching • Thermodynamics driven hybridization • Forms well-known double helix • Molecular Scale • 3.4 Angstrom spacing • 2nm diameter • Synthetic • Specify sequence of bases • Engineer systems [Figure form Pray, Nature Education, 2008] 4

Biomolecules: Chromophores (Fluorophores) Quantum mechanical description of energy levels Fluorescence • Optically active small-molecule • Absorb and emit photons of specific wavelengths • Time to fluoresce follows exponential distribution • Size: ~20-100 atoms 5 Images courtesy of www.invitrogen.com

Biomolecules: Resonance Energy Transfer • Molecular Beacon or Ruler h ν OUT • E.g., detect protein folding h ν IN A B • Resonance Energy Transfer RET (RET) • Closely spaced (1-10nm) • Non-radiative dipole-dipole interaction • Efficiency decays with 6 th power of distance • Efficiency depends on spectral overlap and dipole orientation • Low heat generation (emits far field photon) 6

Why Biomolecules? • Scale in feature size • DNA: 3.4 Angstroms between base pairs • DNA: 2nm diameter double helix • Chromophores: 20-100 atoms • Scale in fabrication • Leverage chemical industry • Engineer systems at low cost and high volume • 1 grad student 8 hours ≈ one month of TSMC Fab 15 throughput • Low Heat Dissipation • Common in Life Sciences • New Domain for computing • Biologically compatible • E.g., computing within a cell 7

How do we use Biomolecules? • Exploit physical properties for Storage 1. Computation 2. Fabrication 3. • Place components (including other biomolecules) • Gates, circuits, systems 8

Biomolecular Storage • Archival Storage P o l y a ; Binary data • DNA base sequence as encoded 01010000 01101111 01101100 01111001 01100001 00111011 Base 3 data 12011 02110 02101 222111 01112 222021 Hu ff man code DNA • Density: 10 9 GB/mm 3 GCGAG TGAGT ATCGA TGCTCT AGAGC ATGTGA nucleotides [Figure form Barnholdt, et al. ASPLOS 2016] • Durability: 100s of years evious Nucleotide • Read Latency: DNA Sequencing • Optical Storage • Photo cleavable link of Chromophore to DNA • Multiple bits w/in diffraction limit • Density: 1000x > blu-ray [Figure from Mottaghi & Dwyer, 2013]. 9

Biomolecular Computation Image from [Zhang & Seelig, Nature Chemistry, Jan 2011] • Specify sequences such that desired hybridization occurs • DNA Computing • Hamiltonian Path, Tile-based computing, • Strand displacement (above) • Attach proteins (molecular recognition) • Molecular Robotics, Synthetic Biology • Chemical Reaction Networks • Careful about different input modes (e.g., concentration of disparate chemicals) 10

Biomolecular Fabrication • Molecular Self-assembly • Molecules self-organize into stable structures • What structures? • What devices? • Nanotubes, nanorods, chormophores, etc. • How does self-assembly affect computer system design? 11

DNA for Structure • Directed Assembly • Functionalize devices, etc. [Dwyer, Trans Nano 2003, Trans VLSI 2004] • DNA Scaffold • Engineered Structures • Origami • Hierarchical • Scale: ~10 14 grids/mL [Rothemund, Nature 2006] A • Can exploit DNA programmability B • “at fabrication computing” [IEEE MICRO 2005] 60 nm 140 nm [Patwardhan, et al. 2004 & 2006; Park et al. 2006; Pistol et al. 2006, ] 20nm 12

DNA Self-Assembled Parallel Processor Node Interconnect Node A Wire [Yan ’03] + (selective Control Processor metallization) B Node • Self-assemble ~ 10 9 - 10 12 simple PE PE 20nm nodes (~10K FETs) • Potential: Tera to Peta-scale • Group many nodes into a SIMD PE computing • Random Graph of Small Scale • PEs connected in logic ring Nodes • Familiar data parallel programming • There will be defects • Scaled CMOS may (does) look similar [Patwardhan, et al., ASPLOS 2006] • How do we perform useful • What about those chromophores? computation? 13

RET-based Stochastic (Probabilistic) Computing RET Circuit Single RET Photon Light Fluorescent Fluorescence PDF P Source Network Molecules Avalanche Detector 𝜇 " 𝜇 # (𝑓 '() * − 𝑓 '() + ) 𝜇 " − 𝜇 # 0 t Resonance Energy Transfer P P • Multi-chromophore structure: phase-type distribution [Wang et al, 2015] . • Can fit most distributions to phase-type distribution [Asmussen et al, 1996] . • New Functional Unit [Wang et al, 2016] . (Wednesday talk…) 14

RET-based Logic • Chromophore types: AND Gate Layout 1. Eval – exciton source eval 2. Out – output, monitored for fluorescence x 1 3. Mediators – connect eval to out R 4. Inputs – x1 and x2 x 2 R Disrupt (no RET) • Excitation represents applying a 1 • • Multistep Cascades out • Energy and Exciton Restoration 0.7 • Biologically compatible Output Fluorescence Probability 0.6 0.5 • Sub-diffraction limit addressable sensing x 1 x 2 0.4 R R [Pistol et al. Small 2010] 0.3 0.2 • Nanoscale Sensor Processor smaller 0.1 than largest known virus [Pistol et al. 0 00 01 10 11 ASPLOS 2009] Inputs NAND Gate Layout & Simulations 15

RET-based Logic Power and Area CMOS (1x) RET- Gate Improvement 15 nm Logic 2 2 AND 294,912 nm 361 nm 816x 2 2 OR 294,912 nm 361 nm 816x 2 2 NAND 196,608 nm 361 nm 544x 2 2 NOR 196,608 nm 361 nm 544x • 15nm CMOS, two-input gates • Power: RET-Logic100x lower than CMOS • Area: at least 500-800x smaller than CMOS • Conservative: Assumes two input gate occupies entire 19nm x 19nm DNA tile • Emit far field photon − > no localized heat generation… 16

The Problem with Exponentials • Desire for more 1.E+83 compute and storage Number of Atoms 1.E+74 • Biomolecular scale 1.E+65 • But...O(n!), O(x n ), etc. 1.E+56 1.E+47 • E.g., storage increases 1.E+38 40%/year 1.E+29 • Not Enough Atoms! 1.E+20 • Earth: • 100 years of storage Storage Earth Universe • 42 node Hamiltonian • Known Universe: • 200 years of storage • 60 node Hamiltonian • Architecture 2030: • Still need algorithms… • Use atoms efficiently 17

Conclusion “ It is not the strongest of the species that survive, nor the most intelligent, but the one most responsive to change .” – Charles Darwin • Technology • May not be a single device technology for the future • Biomolecules Scale in feature size 1. Scale in manufacturing 2. Readily available 3. New Domain for Computing 4. Can exploit physical properties 5. • Interdisciplinary research teams • Scale up technology: from bench to processors (“engineering gap”) • Differing goals/metrics • Need bus driver or shared vision • Publishing can be difficult…but the research is really fun! 18

Duke Nanosystems Overview NANA - General Purpose Dipole Dipole I I I I DNA Self-Assembly Architecture [JETC 2006] Large Scale Networks, Logical [ FNANO 2005, Ang. Chemie G G G G Structure & Defect Isolation 2006, DAC 2006 ] [NANOARCH 2005, 2006, Nanonets O O O O 2006, NanoCom 2009] Nano Devices 3.3 3.T Electronic, photonic, etc. 3.6 3.1 3.7 3.0 3.5 [Nanoletters 2006, IEEE MICRO 2008, 3.4 2. VI 3.H 2.7 6 b D A 2.T 1.2 A Small 2010, IEEE MICRO 2015, ISCA W_S 2.0 b b 2.1 d O1 O1 2.5 c c 1.0 1.1 B B A A 1.H c 1.7 2.2 2016 ] 2.H b b a a 1.3 1.41.5 O3 O3 c c 2.4 g O2 O2 1.6 1.T a b b 2.3 a a d d c c c g g O g b b c c C_IN C_IN c c C_OUT C_OUT a a a d d a a g g SOSA - Data Parallel G a a g g b b C_OUT C_OUT c c R_S a a C_IN C_IN O0 O0 c b b b b c c a a b d d I b b a a g g Architecture [NANOARCH 2006, S S E E S0 S0 S1 S1 Sensing & Processing ASPLOS 2006, JETC 2007, 2009] [ASPLOS 2009. IEEE MICRO 2010, Small 2010] 𝜇 " 𝜇 # (𝑓 '() * − 𝑓 '() + ) 𝜇 " − 𝜇 # NoC Circuit Architecture [NocArc 2014, JETC 2015, [FNANO 2004, IEEE MICRO 2008 0 t ASPLOS 2015] IEEE MICRO 2015, ISCA 2016] Waveguide with DWDM Stochastic Computing Coupler Chromophores [IEEE MICRO 2015, ISCA 2016] Photodetector QD_LED 19