Conference: ACS, New Orleans 2003


Notes from the 2003 ACS Conference in New Orleans


INOR: Poster session 7-10pm

  • 144: Engineering of metalloproteins and novel biocatalysts using native protein scaffolds. J. R. Carey, Y. Lu, et al. Check about SOR (Super Oxide Reductase)
  • 175: Modeling the impace of geometric parameters on the redox potential of blue copper proteins. J. Reglinski, M.K. Taylor.
  • 187: Modeling the active site chemistry of superoxide reductase. J.A. Hallen, D.C. Fox, H.L. Moore, T.C. Brunold, A.T. Fiedler.
    • Need to check more info on SOR.


COMP, Section E, Conv. Center Room 275

Biological Applications of Implicit Solvent Models

  • 8:00a, #300: A “second generation” of generalized Born models for protein simulations. D. Case (Missed the talk)
  • 8:45a, #83: SMx models: New developments and comparison to other electrostatic formalisms. C.J. Cramer, F.J. Luque, et al. (Missed the talk)
  • 9:30a, #84: Dielectric relaxation in proteins with explicit, implicit, and mixed solvent models. T.J. Simonson, G. Archontis, et al. (Missed the talk)
  • 10:15a, #85: Implicit solvent models in biology. Simulation and pH-dependent properties of proteins. E.L. Mehler, S.A. Hassan. (Missed the talk)
  • 11:15a, #86: From implicit solvation to QM: Insights into protein structure. J. Hermans, Y. Vorobjev, H. Hu, G. Buterfoss.
    • ES/IS method: a free energy function for discrimination between folds.
      1. Discriminate in ab initio folding methods
      2. Energetics of allosteric proteins
      3. Discriminate between NMR alternatives
    • Similar to Case and Kollman’s MM/PBSA
    • Later on there have been newer less accurate but faster methods
    • Steps:
      • Make a hole in the solvent (cavity)
      • Put the protein in the solvent (interaction energy)
      • Solvent reorganization (solvation/polarization energy)
    • Self-consistent method
      • G(pol) = Polarization free energy
      • G(cav) = Cavity free energy
    • ES/IS has been tested to driscriminate intentionally misfolded proteins against correct fold (1998)
    • Expanded test: Protein Science, 2001.
    • Packing energy = Lennard-Jones Energy + Energy to deform geometry (enlongation, twisting, compression)
      • Using this packing energy alone had 100% success in discrimation between decoys and good structures
    • Some lessons:
      • Accuracy of ES/IS free energy function is adequate
      • Correlation between free energy and deviation (rmsd) is poor
      • Use a hierarchy of search/selection methods (Baker and Kollman)
      • Packing energy consistently favors the native state by at least 5 kcal/mol: native is the only stable structure
      • Packing determines a unique folded state
    • Ref: Lovell, Word, et al. ‘The penultimate rotamer library’. Proteins, 2000.
    • Ref: Elstner M. Proteins 50, 451-463, 2003
    • Chi-angle distributions in proteins can be modelled w/ a Boltzmann-like (exponential) energy distribution
    • Exponential correlation: P = exp(-bE) –> Most probable distribution (b = 1/(kB * T), kB = Boltzmann constant)
    • Folded proteins have low entropy and low free energy.
    • Components must have low energy, but high energy distortions are probable.
    • Components do not fit perfectly: side chains adjust to improve nonbonded contacts but within reason (large deviations less probable)
  • 12:00p, #87: Implicit and explicit electrostatic models for structure function correlation of biomolecules. A. Warshel.
    • Relationhip between mircroscopic and Macroscopic models
      • Explcit microscopic models have major convergence problems (e.g. PDLD)
      • Simplified microscopic (grid-based)
      • Consistent semi-microscopic (generalized P-B, etc.)
    • Ref: JACS (next week) paper on copper-protein redox differences
    • PDLD: need to run for long periods of time and convergence is not assured
    • A solution is to use PDLD/S-LRA (solvation model w/ Linear Response Approximation)
    • It is applied to the 2 different charges states: when the solvent sees the charge, when it does not see it (pre-organization)
    • That approach minimizes the problems w/ charge contributions
    • The nature of the protein dielectric constant
      • Macroscopic dielectrics calculated purely w/ physical arguments, has nothing to very little to do w/ a “useful” dielectric constant used in protein calculations
      • It depends on what we need/assume:
        • If all is treated explicitely, epsilon = 1
        • If all is induced explicitely, epsilon = 2
        • PDLD/S-LRA, epsilon = 4 or 6
        • In some models epsilon ≤ 0
    • Performance of current models
      • When using ab-initio modelling of proteins, there is a need to average over a big number of configurations
      • Approach: Perform a MM then do a perturbation to the QM state

CINF, Section A, Conv. Center Room 275

Current Status of XML in Chemistry

  • 1:30p, #30: Compressed Chemical Markup Language for compact storage and inventory applications. M. Karthikeyan, D. Uzagare, S. Krishnan.
    • A proposal for CCML - Compressed CML
    • CML can be applied for
      • Publication analysis
      • Data capture
      • Chemical inventory
    • JME-CML : interactive editor that generates CML
    • Size of CML files is a limitation for inventory application
    • Compression approaches:
      • ZIP/LZW algorithms
      • Pattern/Text blocks (repeated text/data/tags)
    • How to search for data in compressed file
    • Proposed to integrate SMILES or equivalent, along w/ coordinate information
    • Dictionary of compressed tags
    • Create individual ids for compressed tags
    • To preserve chemical information, we need to preserve the connectivity table and the coordinate information.
    • Use of template approach (e.g. ACS format, 1999) to preserve connectivity information. This reduces the size of the connectivity representation, by replacing standard/repeated substructures by a symbol.
    • Reason for CCML: encoding into barcode/RF-tags for inventory and structure searches.
    • Proposal of publicly available Dictionary of unique ID for CML tags (IUPAC interest group in CML).
    • IUPAC: Unique identifier for covalently bonded molecules (IChI).
  • 2:00p, #31: New chemical information interchange standards based on CML: A submission for the Object Management Group. M. A. Miller, S.S. Markel, J.C. Esteva, W.L. Sharp.
    • A project from LION Bioscience (the NetGenics division) under the auspices of OMG’s LSR (Life Sciences Research)
    • OMG Standards are used in Model Driven Architecture (MDA): a way of describing a real-world situation.
    • The model:
      • Molecule
        • May contain atoms, bonds
        • May contain other molecules
      • Atom
        • Properties represented as arrays
          • Single valued for complete structures
          • Multiple values for query structures
        • Type defined by chemical element
        • Chirality
        • Coordinates as relationship to Coordinate object
      • Bond
        • Exactly 2 atoms
        • Bond order as an array
    • Future directions:
      • Molecular properties
      • Polymeric structures, zeolites.
      • Combinatorial
      • Reactions
  • 2:30p, #32: Novel applications of XML in chemistry. P. Murray-Rust, H.S. Rzepa.
    • CML2 Schema (
    • XML for creating a “knowledge GRID” for chemistry.
    • The “Chemical Semantic Web”
      • Machines will be able to discover molecules but structure or properties
    • Beilstein indexes about 600 different properties.
    • Properties often are not electronically available: See SELF
    • IUPAC has an activity in XML dictionaries (e.g. CML dictionaries)
    • From the schema code is generated:
      • CML Core
      • CML React
      • CML Spec
      • CML Comp
    • Worldwide Molecular Matrix
  • 3:00p, #33: The family of XML languages in chemistry. H.S. Rzepa, P. Murray-Rust.
    • Requirements for a “knowledge grid”
      • Syntax
      • Validation
      • Semantics
      • Ontologies
      • Discovery mechanisms (Metadata)
      • Rights (“who created/added/revised/owns this?”)
      • Provenance (“can I trust it? is it intact?”)
    • CML 2 is modular, to facilitate extensibility
    • “Interwingling” : intertwining and mingling of data/documents/information.
    • The CML Family Building Kits:
      • CML Core: Molecules
      • STMML: Scientific units
      • CML React: defines reactions
      • CML Spect: spectra data
      • CML Comp: extends CML Core, STMML, CML React to express computational modelling and simulations
      • CML Query: chemical querying and perception
      • CMLX: extensible chemical markup created from the CML family of components
    • Others: GTML = Graph Theory Markup Language



Computers in Chemistry General Contributions

  • 8:30a, #134: Estimation of physicochemical properties of compounds by SPARC S.H. HIlal, L.A. Carreira, S.W. Karichkoff (EPA/UGA)
    • Sparc Performs Automatic Reasoning in Chemistry: program to calc. properties and reactivity based solely on molecular structure.
    • Capabilities: chemical and physical properties.
    • Basically, it calculates a set of core properties, and all others are derived from these
      - Vapor pressure (temp)
      • Activity coefficient (temp, solv)
      • Ionization pKa (temp, pH, solv): microscopic, zwitterionic constant, molecular speciation, isoelectric point
    • Approach: Molecule is broken into functional units, each with intrinsic properties, e.g. p-amino phenol –> -NH2 = Reference point, -Ph- = Resonance, HO- = Electrostatic
    • Intermolecular interaction model: DeltaG(interaction) = DeltaG(dispersion) + DeltaG(induction) + DeltaG(dipole) + DeltaG(H-bond) + DeltaG(mix)
    • Vapor pressure model has been tested by calculating boiling points from -200 C –> 1000 C
      • Non-polars have an error of +/- 3 C
      • Polars have an error of +/- 7.5 C
    • Activity coefficient comparisons (calculated vs experimental): R\^2 = 0.998, RMS = 0.064, N = 491
      • Solubility: R\^2 = 0.987, RMS = 0.40, N = 647
    • Similar good correlations for ionization pKa’s, etc.
    • SPARC accepts as input a SMILE string or a CAS reg number
  • 8:55a, #135: Calculation of tautomeric equilibrium network constants using SPARC S.N. Ayyampalayam, L.A. Carreira (EPA/UGA)
    • SPARC does not do first principle calculations, it is based on mechanistic perturbation models
    • Descriptors: density, polarizability, index of refraction, and H-bonding (alpha/beta)
    • System can calculate tautomeric equilibirum networks, irrespective from the initial input structure
    • SPARC is written in Prolog
    • URL:
  • 9:20a, #136: OptiDesign: Extending optimizable K-dissimilarity selection (OptiSim) for use in combinatorial library design F. Soltanshahi, L. Alkella, R.D. Clark (TRIPOS)
    • OptiDesign: Select diverse representatives from a population
    • In 2D designs (or higher dimensionality), it uses axis pivoting for selection
    • It can do full and sparse matrix designs
    • Maximum dissimilarity approach
    • Multi-block designs also possible
    • Backtrack redesigns
    • User-defined filters, contraints, and evaluation parameters, including bias towards reagents, etc.
    • Ref: J. Mol. Graph. Model. 2000, 18, 404
  • 9:45a, #137: Predictive toxicology using quantum QSAR descriptors from intermediates S. Trohalaki, R. Patcher, K.T. Geiss, J.M. Frazier (Air Force Research Labs)
    • QSARs derived from:
      • neutral descriptors
      • radical descriptors
      • combination of neutral and radical
    • Applied to HAs (halogen aliphatics), which exhibit toxicity, in particular liver toxicity
    • Best correlations when using both neutral and radical descriptors
  • 10:10a, #138: WITHDRAWN
  • 10:35a, #139: WITHDRAWN
  • 11:00a, #140: Web services as applications’ integration tool: QikProp case study V.R. Polyakov, A. Laoul, W.L. Jorgensen (Global LG Informatics/Aventis)
    • QikProp (W. Jorgensen) is a DOS program used to predict biopharmaceutical properties
      • Caculates: solubility, adsorption, distribution, metabolism, and excretion (ADME)
      • Uses 3D descriptors for its internal calculations
      • Input format: SD or MOL (both are MDL formats)


Physical Chemistry Awards Symposium

  • 2:10p, #168: Award Address (ACS Award in Theoretical Chemistry) - Quantum Chemistry in the 21st Century Fritz Schaefer (UGA)
    • Next generation: Explicitly correlated electronic structure methods for molecules
    • “Sub-chemical” accuracy methods: approx. 0.1 kcal/mol, compared to current methods: 1.0 kcal/mol
    • Cusps appear in current orbital methods


ACS Award for Computers in Chemical and Pharmaceutical Research

  • 4:20p, #155: Award Address (ACS Award for Computers in Chemical and Pharmaceutical Research) - Discovery through computation K.N. Houk



Protein Flexibility Large Systems

  • 10:00a, #289: Examining complex processes in macromolecular systems B.R. Brooks, G. Stan, X. Wu, H.L. Woodcock (Nat. Heart, Lung, and Blood Institute.
    • Ref. Stan, Brooks, et al. Bioph. Chem., 203 (chemical character conservation in GroEL)
    • Replica/Path method for searching pathways: QM/MM - CHARMM + Gammes. Position component includes rotation/translation.
    • Nudged Elastic Band (NEB) method. Johnson in “Classical and Quantum Dynamics …” (1997)
    • Approach uses ABNR (Adapted Basis Newton-Rapson) to improve convergence, obtaining a quadratic convergence on rxn pathways (for example)
    • Also discussed a molecular modeling method using low resolution maps (from EM results, res. approx 15 Angstroms)
      • Map object creation from EM data or monomer X-ray data
      • Image density comparison (correlation), including partial correlation and weighting of core features
      • Grid-threading Monte Carlo search for image docking
      • Implemented in latest CHARMM as the EMAP command
      • Ref: J. Struct Bio., 2003
  • 10:40a, #290: Identifying slow (rate limiting?) conformational steps in DNA polymerase beta kinetic pathway: Implications to synthesis efficiency and fidelity. T. Schlick (NYU)
    • MD and TMD (targeted molecular dynamics)
    • Two important methods/approaches amenable for long term MD simulations in biomolecules:
      • Stochastic Path Approach (SPA): PNAS 2002, 99, 10394
      • Transition Path Sampling (TPS): Ann Rev Phys Chem 2002, 53, 291
    • Research tries to elucidate at atomic level the mechanisms of efficiency and fidelity. Identify conformational rearrangements that molecule efficiency and fidelity.
  • 11:20a, #291: Using simulation to garner insight into biomolecular flexibility T.E. Cheatham III (U. of Utah)
    • Ref: “The importance of being floppy” Nat. Struct. Biol.
    • Flexibility —> Function
      • Conformation sampling limitations/solutions
        • Longer simulations
        • Force conform. simulations
    • Need to improve quality of Force Fields too
    • As simulations are running on longer time scales, some deficiencies started to be evident. For example the Cornell FF tends to convert any sequence into an alpha helix in the long run
    • Ref: JACS 2002, 124, 11258 (about prediction and folding simulations)


Protein Flexibility Theory

  • 1:00p, #314: The role of flexibility in protein function: Insights from simulation M. Karplus
    • Nobel Prize could not get visa to enter USA