Processes using High Performance Computing Challenges and - - PowerPoint PPT Presentation

▶
processes using high performance computing
SMART_READER_LITE
LIVE PREVIEW

Processes using High Performance Computing Challenges and - - PowerPoint PPT Presentation

Oct 07, 2023 332 likes •520 views

Computational Microscopy of Biomolecular Processes using High Performance Computing Challenges and Perspectives Divya Nayar Centre for Computational and Data Sciences, Indian Institute of Technology Kharagpur Book cover: T. Schlick Workshop

slide-1
SLIDE 1

Computational Microscopy of Biomolecular Processes using High Performance Computing

Challenges and Perspectives

Workshop on Software Challenges to Exascale Computing 13-14 December 2018, Delhi

Divya Nayar

Centre for Computational and Data Sciences, Indian Institute of Technology Kharagpur

Book cover: T. Schlick

slide-2
SLIDE 2

A living cell environment: Macromolecular crowding

  • Steric interactions
  • Water behaves differently
  • Dynamics affected

2

Large system sizes !

Exascale Representation of a living cell Protein folding-unfolding

~10-100 µm

DNA condensation

slide-3
SLIDE 3

A living cell environment: Macromolecular crowding

Macromolecular crowding needs to be accounted for !

2

Representation of a living cell Current simulation stage: Tera/Petascale

slide-4
SLIDE 4

Breakthroughs: Molecular-level understanding

3

Sanbonmatsu et al. J Struct Biol. 2007, 157, 470–480

Cellular-level systems ?

slide-5
SLIDE 5

Computational Challenges

  • Accurate modelling
  • Large system sizes: N~10 million atoms
  • Long simulation times needed: ~ 100 µsec
  • Large data size generated: ~ 50 TB

Dilute ~ 5X10

4

atoms Crowded ~ 10

7

atoms Current understanding For complete understanding Tera/Petascale Exascale

Needed:

  • Efficient parallel simulations
  • GPU acceleration
  • Making MD packages efficient

4

lFeig et al. J. Phys. Chem. B 2012, 116, 599 lFeig et al. J. Mol. Graph. Model. 2013, 45, 144

David S. Goodsell, the Scripps Research Institute (2016). Http://mgl.scripps.edu/people/goodsell/illustration/mycoplasma

slide-6
SLIDE 6

Molecular dynamics algorithm: Make it efficient !

MD packages (open-source): GROMACS, NAMD, LAMMPS

Input configuration

(interactions between molecules: potential energy functions)

Solving Newton’s equations of motion Update configuration Store snapshots of system

  • ffload

10-100 microsecs Implement latest algorithms like Staggered Mesh Ewald Advanced methods too expensive !

  • Numerous parallel MD simulations
  • GENESIS package for crowded systems

Domain decomposition (load balancing) GPU acceleration CUDA-enabled analysis codes

5

Parallelization schemes: MPI, MPI+OpenMP multi-threading

slide-7
SLIDE 7

Benchmark performance of MD simulations

Intel Xeon E5-2690 CPUs, each with eight 2.9GHz cores ; System size: ~1 million atoms

GENESIS package GROMACS 5.1.2 package

https://www.nvidia.com/en-us/data-center/gpu-accelerated-applications/gromacs

6

Jung et al. WIREs Comput . Mol. Sci. 2015, 5, 310

slide-8
SLIDE 8

Example: Aggregation of α-synuclein protein-Parkinson’s disease

Enabled predicting binding free energy to form Amyloid

α –synuclein monomer

Amyloid aggregate (Parkinson’s disease) Parallel MD simulations of dimers using HPC

  • Dilute solutions !!
  • Only dimers studied
  • ~ 2 µsec (Petascale)
  • GROMACS 2016

7

Ilie, I.M.; Nayar, D. et al. J. Chem. Theory Comput. 2018, 14, 3298

Next step: Realistic cellular environment Challenges to be addressed !

slide-9
SLIDE 9

Centre for Computational and Data Sciences (CCDS) IIT Kharagpur

(Estd. March, 2017)

  • 1.3 Peta-Flop Supercomputing facility : National Supercomputing Mission (NSM).
  • IIT Kharagpur: Nodal Centre for the HR-development activities
  • Interdisciplinary Centre
  • Faculty

working in different HPC application domains: Computational Chemistry/Biology, Material science, Atmospheric Modeling, Computational Fluid Dynamics, Geo-Scientific Computations, Modeling and Mining of Heterogeneous Information Network, Computational Physics, Cryptanalysis, Numerical Mathematics, Computational Mechanics, Non-equilibrium Molecular Dynamics

  • Interdisciplinary teaching for Ph.D./ Master’s students

8

slide-10
SLIDE 10

Thank you for your attention !

Acknowledgements

  • Prof. Nico van der Vegt: Technische Universitaet (TU) Darmstadt, Germany
  • Prof. Wim Briels: University of Twente, Netherlands
  • Dr. Ioana M. Ilie: University of Zurich
  • Dr. Wouter K. den Otter: University of Twente, Netherlands

Computational facility: Lichtenberg HPC Cluster, TU Darmstadt

Organizers of SCEC 2018

9

slide-11
SLIDE 11
slide-12
SLIDE 12

Example 1: Parallel MD simulations protocol

Polymer in aqueous urea solution

Computational resources: Lichtenberg High Performance Computing Cluster, TU Darmstadt

Polymer System System size (atoms)

  • No. of

parallel simulations Total simulation time per concentration Total CPU time (core-hours) Wall clock time per run

  • f 20 ns

(hrs) CPU memory per core PNiPAM 26000 1800 4 μs 648000 9 200 MB PDEA 72000 2000 4 μs 3456000 20 400 MB Total 3800 8 μs ~4.1 million MD package: GROMACS 4.6.7 (MPI enabled, 64-bit) Hardware: Intel(R) Xeon(R) CPU E5-4650 @ 2.70GHz CPU accelerator: avx2

Nayar et al. Phys. Chem. Chem. Phys., 2017, 19, 18156.

Big question: How do cosolvents protect proteins in the cell under extreme conditions ?

urea

8

  • Particle Mesh Ewald: electrostatics
  • Domain decomposition
slide-13
SLIDE 13

NAMD

100M atoms on Jaguar XT5

http://www.ks.uiuc.edu/Training/Workshop/Bremen/lectures/day1/Day1b_MD_intro.key.pdf

slide-14
SLIDE 14

Molecular dynamics (MD) simulations..

Protein structure

Build Simulate Analyze MD packages (open-source): GROMACS, NAMD, LAMMPS Parallelization schemes:

  • MPI
  • MPI+OpenMP multi-threading
slide-15
SLIDE 15

A living cell: Crowded environment !

Representation of a living cell Protein folding-unfolding

~10-100 µm

Water, ions cosolvent, Crowders

3

DNA condensation Microscope

slide-16
SLIDE 16

Our Computational Microscope: Molecular dynamics simulations

Cosolvent Water

High Performance Computing Elucidating molecular mechanisms Molecular simulations

Book cover: Molecular Modeling and Simulation: An Interdisciplinary Guide; Tamar Schlick van der Vegt, N.F.A.; Nayar, D. J. Phys. Chem. B 2017, 121, 9986

2

slide-17
SLIDE 17

Molecular dynamics algorithm: Make it efficient !

MD packages (open-source): GROMACS, NAMD, LAMMPS

Initial configuration Bonded forces Non-bonded forces Integration of equations of motion Update configuration Electrostatic forces (PME) Store samples of configurations

  • ffload

10-100 microsecs Implement latest algorithms like Staggered Mesh Ewald Advanced methods too expensive !

  • Numerous parallel MD simulations
  • GENESIS package for crowded systems

Domain decomposition (load balancing) GPU acceleration CUDA-enabled analysis codes

4

Parallelization schemes: MPI, MPI+OpenMP multi-threading

slide-18
SLIDE 18

Other breakthroughs: Molecular-level understanding

7

Next step: Realistic cellular environment

Sanbonmatsu et al. J Struct Biol. 2007, 157, 470–480

Exascale computing !! Challenges to be addressed