@article {Kolinski2005, title = {Generalized protein structure prediction based on combination of fold-recognition with de novo folding and evaluation of models}, journal = {Proteins}, volume = {61 Suppl. 7}, number = {April}, year = {2005}, month = {jan}, pages = {84{\textendash}90}, abstract = {To predict the tertiary structure of full-length sequences of all targets in CASP6, regardless of their potential category (from easy comparative modeling to fold recognition to apparent new folds) we used a novel combination of two very different approaches developed independently in our laboratories, which ranked quite well in different categories in CASP5. First, the GeneSilico metaserver was used to identify domains, predict secondary structure, and generate fold recognition (FR) alignments, which were converted to full-atom models using the "FRankenstein{\textquoteright}s Monster" approach for comparative modeling (CM) by recombination of protein fragments. Additional models generated "de novo" by fully automated servers were obtained from the CASP website. All these models were evaluated by VERIFY3D, and residues with scores better than 0.2 were used as a source of spatial restraints. Second, a new implementation of the lattice-based protein modeling tool CABS was used to carry out folding guided by the above-mentioned restraints with the Replica Exchange Monte Carlo sampling technique. Decoys generated in the course of simulation were subject to the average linkage hierarchical clustering. For a representative decoy from each cluster, a full-atom model was rebuilt. Finally, five models were selected for submission based on combination of various criteria, including the size, density, and average energy of the corresponding cluster, and the visual evaluation of the full-atom structures and their relationship to the original templates. The combination of FRankenstein and CABS was one of the best-performing algorithms over all categories in CASP6 (it is important to note that our human intervention was very limited, and all steps in our method can be easily automated). We were able to generate a number of very good models, especially in the Comparative Modeling and New Folds categories. Frequently, the best models were closer to the native structure than any of the templates used. The main problem we encountered was in the ranking of the final models (the only step of significant human intervention), due to the insufficient computational power, which precluded the possibility of full-atom refinement and energy-based evaluation.}, keywords = {Algorithms, Computational Biology, Computational Biology: methods, Computer Simulation, Computers, Data Interpretation, Databases, Dimerization, Models, Molecular, Monte Carlo Method, Protein, Protein Conformation, Protein Folding, Protein Structure, Proteomics, Proteomics: methods, Reproducibility of Results, Secondary, Sequence Alignment, Software, Statistical, Tertiary}, issn = {1097-0134}, doi = {10.1002/prot.20723}, url = {http://www.ncbi.nlm.nih.gov/pubmed/16187348}, author = {Andrzej Koli{\'n}ski and Janusz M. Bujnicki} } @article {Kolinski2004, title = {Protein modeling and structure prediction with a reduced representation}, journal = {Acta Biochimica Polonica}, volume = {51}, number = {2}, year = {2004}, month = {jan}, pages = {349{\textendash}71}, abstract = {

Protein modeling could be done on various levels of structural details, from simplified lattice or continuous representations, through high resolution reduced models, employing the united atom representation, to all-atom models of the molecular mechanics. Here I describe a new high resolution reduced model, its force field and applications in the structural proteomics. The model uses a lattice representation with 800 possible orientations of the virtual alpha carbon-alpha carbon bonds. The sampling scheme of the conformational space employs the Replica Exchange Monte Carlo method. Knowledge-based potentials of the force field include: generic protein-like conformational biases, statistical potentials for the short-range conformational propensities, a model of the main chain hydrogen bonds and context-dependent statistical potentials describing the side group interactions. The model is more accurate than the previously designed lattice models and in many applications it is complementary and competitive in respect to the all-atom techniques. The test applications include: the ab initio structure prediction, multitemplate comparative modeling and structure prediction based on sparse experimental data. Especially, the new approach to comparative modeling could be a valuable tool of the structural proteomics. It is shown that the new approach goes beyond the range of applicability of the traditional methods of the protein comparative modeling.

}, keywords = {Amino Acid Sequence, Animals, Carbon, Carbon: chemistry, Crystallography, Databases as Topic, Humans, Hydrogen Bonding, Mathematics, Models, Molecular, Molecular Sequence Data, Protein Conformation, Protein Structure, Proteins, Proteins: chemistry, Proteomics, Proteomics: methods, Tertiary, Theoretical, X-Ray}, issn = {0001-527X}, doi = {035001349}, url = {http://www.ncbi.nlm.nih.gov/pubmed/15218533}, author = {Andrzej Koli{\'n}ski} }