Ned H. Martin

Department of Chemistry and Biochemistry

University of North Carolina Wilmington

Enhancing ComputationalCapabilities in Chemistry (and Grid Computing) at UNCW

Duke University, April 18, 2005Duke University, April 18, 2005

Outline, Part 1

Culture of technology use in Chemistry at UNCW

Grants that provided necessary infrastructure

Phase I of Integrating Modeling into Curriculum: Goals and Strategy

Selective Integration of Modeling into most promising course / instructor combinations to enhance student’s 3D perceptions.

Demonstration of benefits (to win support of faculty).

Phase II of Integrating Modeling into Curriculum

Expand to other courses in chemistry.

Current Efforts / Results / Conclusions

Early Use of Technology at UNCW

1981 – First student microcomputer lab at UNCW (Chemistry)

Spreadsheets, statistics, graphing, word processing.

ProStat statistical analysis/graphing software written by Dick Ward

1986 – Chemical Applications of Microcomputers course

Introduced students to word processing, spreadsheets, and interfacing computers with electronic equipment

1988 – Molecular modeling software obtained

PCModel on pcs,

AMPAC on VAX (gift from Dewar’s group); initially used only in research.

Early Use of Technology at UNCW

1989 – NHM attended NSF Workshop on Molecular Modeling

Week-long workshop at Georgia State University.

1990 – Computational Chemistry courses at NCSC & onlineProvided necessary competence/confidence level for faculty to initiate teaching of computational chemistry methods.

1992 – Introduced Computational Chemistry into Advanced

Organic Chemistry (Physical Organic) course Used computations to illustrate concepts in text; students did not do calculations themselves, just saw results

Grants for Infrastructure

1992 – HyperChem grants in Chemistry and Biochemistry

Software for curriculum development, research.

1993 – NSF Grant for Integrating Molecular Modeling into the Chemistry Curriculum (“Phase I”)

Provided SGI workstation, 8 ‘fast’ pcs, and multiple copies of HyperChem modeling software for chemistry student computer lab and faculty.

Impacted primarily upper level chemistry courses: Organic Chemistry, Advanced (Physical) Organic, Physical Chemistry, Biochemistry, Independent Study.

Grants for Infrastructure…

1994 – NIDA Medication Development Database

Pilot project contract; provided Accord (3D structural database software), student training, led to QSAR projects

1996 – ACS-PRF grant for Modeling NMR Shielding (#1)

Spartan and Gaussian94W software, student support

1997 – NCSC Visualization grant (to NHM)

SGI O2 workstation, AVS visualization software

1998 – NCSC Visualization grant (to MM)

SGI O2 workstation, AVS visualization software

Grants for Infrastructure…

2000 – ACS-PRF grant for Modeling NMR Shielding (#2)

Updated modeling software, student support

2000 – Camille and Henry Dreyfus Grant to Enhance Computational Chemistry Capabilities (“Phase II”)

Impacted courses omitted from 1992 NSF grant: Introductory (General) Chemistry, Inorganic Chemistry, Medicinal Chemistry, and a new course in Computational Chemistry.

Also addressed student research needs, NMR data processing.

Provided SGI workstation, NMR analysis software, 10 fast pcs, multiple copies of Titan.

Grants for Infrastructure…

2001 – Numina Grant for HP Jornadas ` and pocket HyperChem

Allowed student use of computers fro molecular modeling in class; also allowed for instant feedback on student perceptions

2002 – ITSD grant for PocketPCs

Improved in-class devices

2004 – ACS-PRF grant for

Modeling NMR Shielding (#3)

Updated software, student support

Modeling NMR Shielding (#3)

Goals and Strategy, Phase I

Goal (Phase I): To enhance student’s perception of 3D concepts in chemistry:

Stereochemistry; conformations of molecules, and relationship of energy to molecular conformation.

Strategy 1: Selective integration of modeling into the most promising course / instructor combinations (most receptive)

Acceptance by the instructor is key. This sometimes required some time for the “value” of computational chemistry to be recognized.

Training is also needed for those not using modeling in research. This must be repeated each semester for new instructors and TAs.

Goals and Strategy, Phase I…

Strategy 2: Progressively integrate molecular modeling into the chemistry curriculum, starting in sophomore Organic Chemistry

Include some modeling in several courses throughout the curriculum, so that students learn a variety of applications

Verify modeling predictions with experimental results

Teach increasing levels of theory as needed, rather than overloading students with theory to start

Treat molecular modeling as a routine tool, like GC, HPLC, IR, or NMR

Design experiments so that students can “discover” applications of molecular modeling as well as learn its limitations

Specific Objectives, Phase I

1. Develop computational exercises with experimentally verifiable results for selected courses.

Predicting the major alkene isomer resulting from dehydration of an alcohol. (Organic Chemistry)

Base pair H-bonding stabilizes DNA. (Biochemistry)

2. Test student’s perception/knowledge level before and after modeling was introduced to determine the effect of the curriculum change.

3. Provide adequate and ongoing instructional / tutorial support for students and faculty/TAs.

Gain support and confidence of faculty.

Intro. to Molecular Mechanics

Organic Chemistry students learn the basics of molecular mechanics

Create models of structures, perform energy minimizations

Measure bond lengths, bond angles, and dihedral angles

Construct model of axial methylcyclohexane using “ideal” bond lengths and bond angles; measure these.

Perform energy minimizations and observe how the molecule adjusts its structure to minimize its energy; measure the same bond lengths and bond angles after energy minimization.

109.5109.5°°

112.2112.2°°

Organic Chemistry Experiment

Compute the energies of the isomeric carbocations that arise from acid-catalyzed dehydration of an alcohol.

Sayed, Y.; Ahlmark, C. A.;Sayed, Y.; Ahlmark, C. A.; Martin, N. H. Martin, N. H. J. Chem. Educ. J. Chem. Educ. 19891989, , 6666, 174-175. , 174-175.

C

CH3

CH3

CH3

CH

OH

CH3

C

CH3

CH3

CH3

CH CH3

+ H2OH

CCH3

CH3

CH CH3

CH3

(2º carbocation)(2º carbocation)

(3º carbocation)(3º carbocation)

methide shiftmethide shift

Organic Chemistry Experiment…

Computation shows that the rearranged 3º carbocation is much lower in energy; it can lose H+ to form either of two alkenes; the one that predominates according to GLC analysis is the lower energy alkene, also shown by calculation.

Martin, N. H. Martin, N. H. J. Chem. Educ.J. Chem. Educ. 19981998, , 7575, 241-243., 241-243.

CCH3

CH3

CH CH3

CH3

C CCH3

CH3

CH3

CH3

and

C CHCH3

CH3

CH2

CH3

major product; lower energy

minor product; higher energy

Biochemistry Experiment

Students model pairs of DNA bases (C-G, A-T, as well as others) using semi-empirical MO theory; they determine the strength of the H-bonds; C-G (top, which forms three H-bonds), has the greatest stabilization due to H-bonding; A-T (bottom) forms only 2 H-bonds.

Biochemistry Experiment…

A plot of the mol % C-G vs. the literature value of ‘melting temperatures’ (temperature at which the helix unravels) of various DNA samples is linear.

This demonstrates the effect of H-bonding on stabilizing the double helix.

Martin, N. H., Burgess, S. K., Connelly, T. L., Reynolds, W. R.; Martin, N. H., Burgess, S. K., Connelly, T. L., Reynolds, W. R.; Spiro, L. D. Spiro, L. D. Biochemical EducationBiochemical Education 19961996, , 24(4)24(4), 230-231., 230-231.

Specific Objectives, Phase II

1. Develop computational exercises with experimentally verifiable results for additional selected courses.

Shapes of simple molecules; VSEPR rule ‘verification’. (General Chemistry)

Orbital shapes and energies; transition metal complexes. (Inorganic Chemistry)

Relating electrostatic energy to stability in carbocations. (Physical Organic Chemistry)

2. Develop new Computational Chemistry course.

3. Provide ongoing instructional / tutorial support for students and faculty/TAs.

General Chemistry

Hand-held Dell Axim PocketPCs (left) runing HyperChem provide students with in-class opportunity to view and rotate 3D structures, measure bond angles, and examine molecular shapes and resulting properties, such as polarity.

Bond angle calculation

Experimental group used HyperChem to rotate molecules and measure bond angles

Control group used the PocketPCs to view structures in color, but with no rotation capability

Sample Quiz Questions

Gas Law QuestionGas Law QuestionVSEPR QuestionsVSEPR Questions

Test Results

(control)(control)

Inorganic Chemistry

HP Jornadas or PocketPCs and HyperChem are used in Inorganic (CHM 445) lecture to visualize molecular orbital splitting, see the shapes of molecular orbitals and their energy levels, and calculate bond stretching frequencies of CO before and after complexation with a metal.

Inorganic Chemistry…

Students compute the energies of the molecular orbitals of BH3 (top) and then visualize them (bottom) to assess Lewis acid properties.

Physical Organic Chemistry

Students use Jornadas or PocketPCs and HyperChem during lecture to examine various topics as they are discussed, including:

MO calculations of molecular geometry, bond orders, atomic charges, and hybridization.

Visualization of symmetry properties of molecules:

Calculation and visualization of steric effects in substituted cyclohexanes.

Students also do computational projects outside of class using HyperChem on pcs in the computer lab.

Computational Chemistry course

Computational Chemistry…

New course in 2002, 2 lecture & 2 computer lab hours/wk

http://www.uncwil.edu/chem/molecularmodeling

Covers the basic theoretical background of several computational methods: molecular mechanics, quantum mechanics, density functional theory, molecular dynamics.

Provides computer lab exercises in model building, energy minimization,conformation searching, transition state modeling, reaction pathway modeling, visualization of results and molecular property calculations (NMR).

Introduces solvent effects, QSAR, modeling biomolecules, UNIX language, grid computing.

Comp. Chem Syllabus

Introduction to computational chemistry (overview of capabilities, relative cpu time, limitations and applications of various methods)

Molecular mechanics (components of force fields, file types, atom types, successes and limitations; caveats about minimum energy structure)

LAB 1. Building and optimizing structures in Titan (model building, rendering modes, measurements)

Molecular orbital theory, part 1 (history, levels of MO theory, SEMO methods, computational results)

LAB 2. Manual conformation searching methods

Molecular orbital theory, part 2 (ab initio MO theory, basis sets, correlated methods, effect of choice of method/basis set on cpu time)

LAB 3. Automated conformation searching

Calculating molecular properties (energy derivatives, UV-Vis, NMR, freq.)

Comp. Chem Syllabus

LAB 4. Ring strain in cycloalkanes; isodesmic reactions

Potential energy surfaces; optimization methods; reaction path following (gradient, stationary points, saddle point, minimization algorithms, TS modeling, frequency calculation, rxn. pathway calc.)

LAB 5. Modeling a reaction pathway; the pinacol rearrangement (locating a TS; frequency calculation;

Computing charges on atoms (Mulliken, natural bond order, AIM, MK and CHELPG charges; best fit to NMR data; electrostatic effects on carbocation stabilization and conformation)

LAB 6. Stability of alkenes and carbocations

Solvation effects; hybrid (QM/MM) methods (explicit, continuum and hybrid models; ONIUM method; hybrid MM/QM methods)

LAB 7. Basicity of amines (electrostatic potential mapped on electron density isosurface; modeling solvent effects)

LAB 8. Modeling bromonium ion intermediates (LUMO)

Comp. Chem Syllabus

Density functional theory (guest lecturer Lee Bartolotti, ECU)

LAB 9. Endo/Exo Selectivity in Diels-Alder Cycloadditions (kinetic vs thermodynamic control)

Grid Computing; UNIX operating system; Remote computing; Gaussian 03; GridNexus; NMR calculations of classical vs. non-classical carbocations

LAB 10. Modeling the Relative Acidities of Substituted Phenols (npa charges, electrostatic potential mapped on electron density isosurface)

Quantitative Structure-Activity Relationships (QSAR)

LAB 11. NMR shift and charge calculations using Gaussian 03 on a Linux cluster; Introduction to Grid computing via GridNexus (file formats and their interconversion)

WWW computational chemistry resources; modeling biomolecules (special visualization methods)

Summary and Conclusions, Part 1

Computational applications have been integrated throughout the chemistry curriculum at UNCW.

The process requires interested / convinced faculty.

Ongoing training of faculty and TAs is critical.

We found that to be most effective, computer exercises should be verified by laboratory results.

Integration into multiple courses and all levels (freshman through senior level) is critical in order to demonstrate to students the general applicability of computational methods.

Part 2. Grid Computing

Rationale for Grid Computing

0

2000

4000

6000

8000

10000

12000

14000

1960 1970 1980 1990 2000 2010

MIPS

Series2ts/104

The recent The recent proliferationproliferation* of * of fastfast, , interconnectedinterconnected underutilized cpusunderutilized cpus

* over 150,000,000 pcs are sold each year!* over 150,000,000 pcs are sold each year!

Grid Computing

A computing Grid is analogous to an electrical power grid. The user simply “taps” into the resource (with permission), but is usually unaware of the origin of the resource.

Grid Computing at UNCW

Current efforts by a group of UNCW computer science faculty and undergraduate students, plus faculty and students in several “application areas” are focused on developing a graphical user interface (GUI) called

GridNexus serves as a front-end to simplify data manipulations, searching or calculations of various types performed on remote computers over a Grid.

This project has grant support from the UNC Office of the President

GridNexus

GridNexus is based on JXPL, a new graphical programming language developed by UNCW computer science faculty and students.

GridNexus allows users to link modules that perform various operations into a usable ‘workflow’, then save these for later use.

Once a ‘workflow’ has been created, one only need to specify the path/filename of the data set to be operated on and the path/filename for the output file.

This greatly simplifies repetitive operations, and takes much of the mystery out of computing for non-computer science users.

File Interconversion in GridNexus

One of the limitations of most computational chemistry software packages is that they do not read or write many different (proprietary) file types, so it is difficult to transfer data from one program to another.

GridNexus allows users to input some of the most common types of geometry specification, such as .pdb (.ent) and .mol files, and use a default set of options (or select from a list) to write a Gaussian input (.dat) file.

GridNexus also allows the user to orient a molecule in a specified way in Cartesian coordinates.

Gaussian 03 under GridNexus

Functions can be Functions can be selected from lists at selected from lists at the top left, dragged the top left, dragged onto the workspace onto the workspace and joined.and joined.

The entire workflow can be hidden The entire workflow can be hidden in a single multifunction boxin a single multifunction box

Gaussian 03 under GridNexus

Submitting a Gaussian job can be as simple as selecting the input file name (from a variety of file types) and the desired output file name.

Molecule Orientation in GridNexus

One module allows a molecule to be oriented in Cartesian space in a specified way, then writes a proper Gaussian03 input file.

Gaussian 03 Input File

%chk=tmp/martinn/phenanthreneNH2.chk

# HF/6-31G(d,p) opt freq

phenanthreneNH2

0 1

H -1.963715 -3.198017 1.280991

C -1.127512 -2.730904 0.750482

H -0.184242 -4.593909 0.244859

C -0.149560 -3.501921 0.166986

C 0.000000 -0.715690 0.000000

N 0.000000 0.715690 0.000000

C 0.908090 -2.892498 -0.536779

C -1.036579 -1.338948 0.691052

C 0.971979 -1.491079 -0.702775

C 1.943981 -3.742718 -1.057698

H -1.800364 -0.744862 1.210005

H 1.238823 1.070292 -1.769705

C 2.993024 -3.223318 -1.730309

(etc.)

Note C & N along the Y axis, the midpoint of their bond at the origin

What’s next for GridNexus?

Develop more “filters” to transform data.

Enhance the graphics for appearance and usability.

Include more software applications.

Extend Grid services to other disciplines.

Include industry and businesses as users and developers.

Add more computational nodes to the Grid. The goal is to include all NC institutions of higher learning

Acknowledgements

NSF

ACS-PRF

HyperCube, Inc.

Pearson Education Foundation

Camille and Henry Dreyfus Foundation

UNCW: Department of Chemistry and Biochemistry, College of Arts and Sciences, Division of Academic Affairs, and Information Technology Systems Division (ITSD)

(former) North Carolina Supercomputing Center

UNC-Office of the President