SLIDE 1
Bacteria are among the earliest forms of life that appeared
- n Earth billions of years ago. Studied since 1676…
Bacteria are of immense importance because of their extreme flexibility, capacity for rapid growth and reproduction.
2
Fighting Infections and Antimicrobial Resistance Through - - PowerPoint PPT Presentation
Fighting Infections and Antimicrobial Resistance Through - - PowerPoint PPT Presentation
Fighting Infections and Antimicrobial Resistance Through GPU-Accelerated In Silico Models Radu Marculescu Carnegie Mellon University GTC, April 5, 2016 Bacteria are among the earliest forms of life that appeared on Earth billions of years ago.
SLIDE 2
SLIDE 3
The fight against bacterial infection represents one of the big challenges of modern medicine.
4
Focus on intercellular communication to understand behavior at population-level
SLIDE 4
Mathematical characterization of the collective dynamics
- f such bacterial swarms represents a major challenge
5
Having a quantitative model of pathogen interactions is crucial for understanding how to fight them.
SLIDE 5
6
This presentation focuses on the “network” paradigm and ways to engineer communication
Silicon networks On-chip communication
Bacteria form a “social network” to coordinate their behavior, build the biofilm and resist treatment.
Biological networks Cellular communication
SLIDE 6
A New Frontier – Quorum Sensing. QS is a density- dependent collective behavior that enables communication
lasR lasR lasI lasI LasR LasR LasR LasR LasI LasI
+ +
AI
+
LasA LasB ToxA AprA XcpP XcpR virulence virulence
QS upregulates multiple virulence genes
Gram-negative bacteria use largely homologous QS networks, where the AIs are detected and regulated via genetic circuits.
7 [G. Wei, Marculescu, NanoCom, 2014]
SLIDE 7
Bacterial biofilm: Use Twitter-like metaphor to explain participants and network formation
(a) Intracellular network (b) Intercellular network (c) Signal molecule (molecular tweet) EPS/Virulence (d) Signal-negative cheater Signal-blind cheater Wild-type cooperator
8 @bacteria: generate public goods (EPS)
- r virulence
The structure and dynamics of the inter-cellular communication network is heavily influenced by its environment
[G. Wei et al., ACM BCB, 2015]
SLIDE 8
Bacterial population dynamics is a complex problem. Direct wet-lab experimentation is costly and often impractical.
Experimenting with efficient GPU kernels using NVIDIA Thrust allows to achieve 100x acceleration with GTX980 GPUs.
Biomass layer Diffusion layer Bulk layer Nutrition Bacteria
Threshold reached? Yes No Initial state Physical Epoch Biological Epoch
10
74 82 90 98 106 114 122 50 100 150 Number of Particles (x1000) Iterations 0.1 0.01 0.001
High-Accuracy = More iterations
SLIDE 9
- Arrange the data by physical location
- thrust::sort_by_key
- For each agent
- Examine adjacent grid space
- Utilize __shared__ cuda memory
- Determine movement vectors between
agent and all other cells
- Move agent with movement vector
- Repeat until system approaches steady
state
- thrust::transform_reduce to get total
movement in the system
GPUs allow us to maintain high accuracy during the physical interactions within reasonable time constraints (hours vs days)
Parallelize all agents on the GPU. Execute many of these physical interactions concurrently
SLIDE 10
S4 – 1/3WT, 1/3SB, 1/3SN: Bacteria communication enables social intelligence (“molecular tweeting”)
(a) (b) (c)
Link
(300x300x500μm3) Red: WT Blue: SB Yellow: SN Green: EPS
t=0.2 hrs t=10 hrs t=24 hrs
SLIDE 11
Dynamics of network evolution
(a) (b) (c) (d) (e) (g) (h) (f) (i)
SLIDE 12
Effect of QSIs
10:30 12:47 14:00
SLIDE 13
22
This presentation focuses on the “network” paradigm and ways to engineer communication
Silicon networks On-chip communication
Bacteria form a “social network” to coordinate their behavior, build the biofilm and resist treatment.
Biological networks Cellular communication
SLIDE 14
Parallel programs enforce correctness through communication (i.e., synch primitives and cache coherence)
Physical interactions are computationally-intensive and require new approaches to heterogeneous parallel computing.
23 [V. Balaji et al., WAX 2016]
SLIDE 15
…
Novel distributed CPU-GPU architecture can take advantage of the data-locality between GPU cores
Reduce global access by using low-cost data accesses from adjacent GPUs on an integrated distributed system
… … … …
Shared Mem Shared Mem Shared Mem Shared Mem
SM SM SM SM SM SM SM SM
L1 $ L1 $ L1 $ L1 $
Global Memory Low-Latency Nearby Memory Access High-Latency Memory Access
SLIDE 16
How about future architectures? Can we design an on-chip Computational Microscope?
CPU/GPU Modules
L2 L2
CPU/GPU Cluster_N CPU/GPU Cluster_1
…
wired link wireless link NI: network interface CPU/GPU L1 CPU/GPU CPU/GPU Router
To L2
To other routers CPU/GPU Cluster
NI
L1
NI
L1
NI
L1
… …
L1 concentration
Efficient Wireless NoC Architectures
CPU/GPU L1
NI
L1
0.4 0.6 0.8 1 1.2
CANNEAL FFT LU RADIX VIPS WATER
EDP w.r.t. Traditional Arch. VFI-mesh VFI-mSWNoC
NVFI Mesh
[R. Kim et al., IEEE TC 2015]
SLIDE 17
- comp. models
wet lab
SoC-based Computational Microscope
48-72 hrs to obtain in lab
in silico prediction of gene/protein families & molecular pathways
Biofilm
Intra-cellular Dynamics Need ~ 10 ODEs per bacterium to characterize intra- cell processes Biofilm Groups of cells (103) ~ 104 interaction
- eqns. (chemical &
physical processes) Need ~ 107-1010 eqns. for biofilm dynamics (chemical & physical processes) Pairwise molec. comparison ~104-106 integer updates per pair Microbiome Molecular Interactions Global molec. interaction # Sequences: ~106-1010 # Graph edge trav-
- ersals: ~1010-1015
Population-scale interaction # species per micro- biome: ~102-104 # individual cells: ~1010-1013 Inter-cellular network Bacterium (single cell) Population of cells (106-109) Molecule Protein
1-2 hrs to simulate
There is a lot more to explore…
SLIDE 18
Summary
- GPU-based simulation is crucial for obtaining faster than real-
time results. Only about 1-2 hrs are needed to simulate a scenario that may take 48-72 hours in the lab
- Network metrics correlate well with biofilm metrics and can
explain population level behaviors of biological significance. Solid basis for studying socio-microbiology…
- Inter-cellular communication is crucial for understanding
biological systems (which are or should become of interest to computer engineers…) Contributors (in no particular order…)
- G. Wei (CMU), I. Cazan (CMU), L. Chieh (CMU), K. Bhardwaj (CMU), C. Walsh