Personal Perspectives on Virtual Laboratory of Computational Chemistry

1
Personal Perspectives on Virtual Laboratory of
Computational Chemistry

• Shubin Liu, Ph.D.
• Academic Technology Networks
• University of North Carolina
• Chapel Hill, North Carolina, USA

2
Agenda

• Self-Introduction
• Research Interests
• Hardware and Software in Use
• Personal Suggestions
• What Can I Contribute?
• Technical Issues and Concerns

3
Brief Self-Introduction

• Education
• Ph.D. University of North Carolina, 1996, Advisor
  Robert G. Parr
• M.Sc. Lanzhou University, 1988, Advisor Zhouren
  Peng
• B.Sc. Hunan Normal University, 1985, Major
  Chemistry
• Professional Experience
• Postdoc. Duke Universtiy, 1998-2000, Advisor
  Weitao Yang
• Postdoc. University of North Carolina, 1996-1998,
  Advisor R.G. Parr
• Present Position
• Title Senior Computational Scientist
• High Performance Computing Group, ITS, UNC, since
  September 2000
• Permanent staff position
• Duties
• Computational Chemistry software installation,
  test, and support for entire campus
• Computational Chemistry research and
  collaboration with world-wide researchers

4
Research Interests

• Density Functional Theory
• Theoretical, conceptual and methodological
  development
• Applications to biological and material systems
• Structure Theory of Macromolecules
• New theory of molecular structure and change
• Structural bioinformatics

5
Ongoing CPU/IO-Demanding Projects

• Cytochrome human P450 protein mechanism
• System 100 atoms of QM core, 3500 atoms of MM
  environment
• Methods DFT, QM/MM
• Software Gaussian, GAMESS
• Quantum Molecular Dynamics
• System 100 atoms, 20-100 ps of dynamics
• Software NWChem/ECCE
• Structure Bioinformatics
• Self-developed algorithm/codes
• Huge database (gt10GB raw data from one cycle)
• Both CPU and I/O intensive

6
Active Site of P450cam Protein
7
Ongoing CPU/IO-Demanding Projects

• Cytochrome human P450 protein mechanism
• System 100 atoms of QM core, 3500 atoms of MM
  environment
• Methods DFT, QM/MM
• Software Gaussian, GAMESS
• Quantum Molecular Dynamics
• System 100 atoms, 20-100 ps of dynamics
• Software NWChem/ECCE
• Structure Bioinformatics
• Self-developed algorithm/codes
• Huge database (gt10GB raw data from one cycle)
• Both CPU and I/O intensive

8
QMD Chloride in Water Cluster
9
Ongoing CPU/IO-Demanding Projects

• Cytochrome human P450 protein mechanism
• System 100 atoms of QM core, 3500 atoms of MM
  environment
• Methods DFT, QM/MM
• Software Gaussian, GAMESS
• Quantum Molecular Dynamics
• System 100 atoms, 20-100 ps of dynamics
• Software NWChem/ECCE
• Structure Bioinformatics
• Self-developed algorithms/codes
• Huge database (gt10GB raw data from one cycle)
• Both CPU and I/O intensive

10
Structure Bioinformatics Helix Propensity
Huge database, self-developed algorithm, both CPU
I/O intensive. Unpublished results.
11
Hardware in Use

• SGI Origin 3800, 500MHz, 64 CPU, 128 GB memory
• IBM P690 (Regatta), 1.7GHz, 32 CPU, 128 GB memory
• LINUX cluster, 1.4-2.8GHz, 352 CPU, 1.5GB
  memory/CPU AMD Athlon and IBM BladeCenter
• Shared scratch disk space over 10 TB
• All machines for scientific/research computing
• Shared by all users on campus

12
Software in Use

• Quantum Chemistry
• Gaussian
• NWChem
• GAMESS
• ADF/Band
• Cerius2
• SPARTAN
• Molpro
• CPMD
• ABINIT

• Molecular Dynamics
• AMBER
• CHARMM
• Gromacs
• Tinker
• Sybyl
• InsightII
• MacroModel
• Quanta
• O

13
Software in Use

• Visualization
• GaussView
• ECCE (with NWChem)
• Cerius2
• Sybyl
• InsightII
• Molscript
• Spartan
• Molden

• Miscellaneous
• Babel (OpenBabel)
• Cambridge database
• NBO/NBOView
• Blast/Pise
• BLAS/LAPACK
• IMSL
• Matlab, Mathematica
• etc.

14
Suggestions Operational

• CPU-time allocation based on proposals
• A Scientific Steering Committee reviews proposals
• Renewal yearly on the basis of progress report
• Different kinds of account
• Training account
• Student account
• Research/member account
• Mandatory prior training (system, etc) for any
  new user

15
Suggestions Scientific

• Install a selective list of computational
  chemistry software at system level and shared by
  all CC users
• Categories QM, MD, Visualization, utility
• Priority given to parallel codes/packages
• A few use-friendly GUIs for each category
• Symposiums
• have peers get together often and invite
  experts to come and talk
• User forums
• display what users achieve, harvest feedbacks
  from users, and provide opportunity for users and
  system people to have direct talks.

16
Suggestions Training

• Training and short courses
• For new users, students, professionals, etc.
• Format Online and/or onsite
• Contents
• System usage, job submission, HPC technology
  (MPI, OpenMP, etc.)
• Programming languages (FORTRAN 77/90/95, C)
• Comp Chem general introduction, package
  introduction, etc.
• Summer schools with credit points
• Onsite, short-term, intensive, focused, hands-on,
  project-oriented

17
Suggestions Miscellaneous

• List server -- for Lab members
• Website to general public
• User group of all Comp Chem users

18
My Purpose to be a Member

• To provide voluntary service, based on my years
  of HPC experience, to make the Lab more effective
  and efficient
• To get to know peers within our country
• To make use of the great facility and resources
  for own research

19
What Can I Contribute?

• Software
• Selection of a list of CC software
• Free, commercial
• Serial, parallel
• QM, MD, visualization, utility, etc
• Installation, testing, documentation and
  benchmarking
• Consultation
• User (front-end) support
• Technical (back-end) support
• Scientific steering committee
• Training/Workshop/Summer School
• Symposium

20
Technical Issues and Concerns

• How to share data within a group?
• How to guarantee data privacy if needed?
• How much is the disk space quota for a regular
  user?
• Whats the limit of CPU usage per user?
• Is there any permanent mass storage?
• How big is the scratch space, where there is no
  limit to store raw data and to run jobs? How long
  can raw data be kept on the scratch space?
• Is there any backup plan? If yes, what parts of
  disk spaces are to be backed up?
• Any packages/compilers/libraries (BLAS/LAPACK/
  IMSL/etc) installed by system people that are
  available for all users?

21
QUESTIONS???
Please contact me at shubin_at_unc.edu.
THANK YOU!