Laboratory of Theory of Biopolymers, Faculty of Chemistry
Wersja Polska
Laboratory of Theory of Biopolymers Laboratory of Theory of Biopolymers
Enter the e-mail address on which you wish to be notified about news on the site.

Research
Our scientific interests include:
  • computer modeling of proteins and polymers: structure, dynamics and thermodynamics
  • hierarchical approach to protein folding
  • in silico prediction of protein structure: from comparative modeling to de novo folding
  • bioinformatics and biological statistics
  • computer aided drug design
  • modeling and predicting of biomacromolecular interactions: prediction of protein function
  • development of software for large scale molecular modeling and computational analysis of experimental data on biomacromolecules
    Diagram illustrating relations between various areas of our research. Image designed by Sebastian Kmiecik

    Selected research topics

    Structure prediction

    Over the last several years we developed a series of high resolution lattice models that proven to be very powerful tools for modeling protein structure, dynamics and thermodynamics. Recently built CABS model (CABS download page, the acronym comes from the names of united atoms employed in the reduced representation of polypeptides: CA, cB and Side groups) could be effectively used for high resolution comparative modeling as well as for purely de novo folding of small globular proteins (movie of protein G folding).

    The lattice-based CABS has 800 allowed Cα-Cα vectors, which eliminates all possible adverse lattice artifacts. The positions of up to two side chain united atoms are average rotameric states for a given residue type and a given conformation of the main chain fragment. The lattice spacing is equal to 0.61Å. The resulting average cRMSD for the Cα-trace of PDB structures fitted to this lattice is in the range of 0.35Å. The actual resolution of the model is in the range of 1-2 Å depending on protein size, which makes it an appropriate tool for fold refinement in comparative modeling, and de novo fold assembly. The knowledge-based force field of CABS consists of several potentials of mean force derived from careful statistical analysis of structural regularities seen in known protein structures.
    The CABS-based hierarchical methodology for protein structure prediction proved to be very efficient during the last CASP6 (Critical Assessment of Protein Structure Prediction), where predictions submitted by group Kolinski-Bujnicki was ranked as the second to the best among over the 200 groups participating in the experiment:

    CASP6 average scoring, all categories (New Folds,
    Fold Recognition,     Comparative Modeling):
    1. Ginalski (ICM, Poland)
    2. Kolinski & Bujnicki (UW-IIMCB, Poland)
    3. Baker (USA)
    4. Skolnick_Zhang (USA)
    5. GeneSilico (IIMCB, Poland)
    Description of CABS performance during CASP6: A. Kolinski, J.M. Bujnicki, Proteins 2005, 61(S7):84-97 available on-line.

    Schematic representation of the CABS model


    Prediction of target T0223 from the CASP6 experiment (comparative modeling).

    Protein length 206 aa. Templates: 1noxA, 1bkjA. Three initial threading models: 3.2 - 4.5 Å from the native structure. After simulations: 3.0 Å from the native, domains: 3.1 and 2.8 Å.


    Prediction of target T0201 from the CASP6 experiment (de novo folding of a new fold).

    Full length protein 5.1 Å from the native, fragment 8-84 2.6 Å from the native.


    Multiscale Modeling Tools for Protein Structure and Dynamics Simulations

    Recently we developed a suite of tools for a multiscale computational approach to protein structure prediction and modeling protein dynamics on biologically important time-scales. Our recent studies show, that reduced modeling emploing knowledge-based, statistical potentials, derived from known structures, is applicable in simulations of protein denatured state and protein folding mechanism studies. More at our multiscale modeling page.

    Development of software and computational methods for bioinformatics and molecular modeling

    We developed a number of complete modeling tools for polymers and biopolymers, including the previously mentioned CABS and its continuous-space relative REFINER. The most recent versions of these programs enable flexible docking of peptides to proteins and flexible assembly of protein aggregates. New efficient versions of the Monte Carlo sampling techniques have been developed for these models.

    Docking of short peptide to the vitamin D receptor


    De novo assembly of Rop dimer

    A number of novel bioinformatics tools were also designed recently enabling easy construction of automated pipelines for large scale structure prediction, derivation of protein-specific knowledge based potentials for threading, structure modeling and model selection and structure clustering and analysis. Almost all software designed in our group is freely available from this page or upon request from the authors (for academic purposes only)

    BioShell service.


    Idealized models of proteins

    The protein model we adopt is a face-centered-cubic lattice chain, with the chain beads representing the polypeptide amino acid units. Each amino acid residue is characterized by two fundamental properties: its hydrophobicity (that dictates the character of the binary interactions) and its secondary structure propensity (that encodes the tendency to adopt a specific rotational-isomeric state of a chain fragment). Such an interplay between the short- and long-range interactions leads to cooperative collapse transitions. We provided quantitative arguments that the existence of both types of interactions is actually a necessary condition for protein-like behavior. It has been demonstrated for three types of protein motives: a β-sheet protein, α-helical bundle and a mixed α/β motif. The sampling and thermodynamics analysis employed a multicopy Monte Carlo techniques.

    Density of states shows cooperative two-state folding equilibrium.

    Density of states shows cooperative two-state folding equilibrium.

    In silico modeling of prion propagation

    There is a hypothesis that dangerous diseases such as BSE, Creutzfeldt-Jakob, Alzheimer, fatal familial insomnia and several other are induced by propagation of wrong or misfolded conformations of some vital proteins. If for some reason the misfolded conformations were acquired by many such protein molecules this might lead to a "conformational" disease of the organism. We proposed theoretical model of the molecular mechanism of such a conformational disease, in which a metastable (or misfolded) form of a protein induces a similar misfolding of another protein molecule. First, a number of amino-acid sequences composed of 32 amino acids have been designed that fold rapidly into a well-defined native-like α-helical conformation. From a large number of such sequences a subset of 14 had a specific feature of their energy landscape - a well-defined local energy minimum (higher than the global minimum for the α-helical fold) corresponding to β-type structure. Only one of these 14 sequences exhibited a strong autocatalytic tendency to form a β-sheet dimer capable of further propagation of protofibril-like structure. Simulations were done using a reduced, although of high resolution, protein model (REFINER) and the Replica Exchange Monte Carlo sampling procedure.


    Prion propagation, a misfolded protein induces misfolding.


    Example of the β-sheet trimer obtained in three chain dynamics.

    Initial geometries of molecules with frozen β-sheet and trimer three views are shown.

    Movies from the simulation data

    Biomedical Computation Review, Fall 2005, page 4

    Modeling & Docking

    Our group has extensive experience in comparative modeling, ligand docking and protein-protein docking. We use our own software, as well as standard packages: Modeller, Sybyl and others.

    Modeling with experimental data

    Combining theoretical approach with practical experiments accelerates determination of protein structure without limiting the model resolution. We successfully use the NMR data together with the CABS protein model and the NMRPipe software.

    Collaboration

    Collaboration with experimental groups in interesting experimental/theoretical studies is always welcome.
Warsaw University
Warsaw University
www.uw.edu.pl
Contact us
Laboratory of Theory of Biopolymers, Faculty of Chemistry UW
ul. Pasteura 1, 02-093 Warszawa
tel.: (+48) 22 8220211
fax: (+48) 22 8220221, (+48) 22 8225996
design by VENTI Copyright © - 2005 Laboratory of Theory of Biopolymers, Faculty of Chemistry