ABSTRACT The protein structure prediction problem continues to elude scientists. Even though many... more ABSTRACT The protein structure prediction problem continues to elude scientists. Even though many new methods have been introduced, certain classes of prediction targets such as free modeling targets remain a challenge based on blind predictions in the several previous Critical Assessment of Techniques for Protein Structure Prediction (CASP) experiments [1]. To meet this challenge, a large-scale collaborative effort called WeFold was undertaken by thirteen labs, each with their own specialties and approaches in addressing the problem. In this talk, we will present the different methods or branches collaboratively designed and tested during the WeFold experiment, as well as their predictive ability, outcomes, and lessons learned. Independent branches involved in the collaborative effort yielded several high-ranking predictions among all group and method submissions in CASP10 for human, free modeling (template free), and refinement targets. Remarkably, two WeFold methods were able to produce the very best predictions in the refinement category in two different targets from a starting model for several different accuracy metrics. Contributions were made by junior and seasoned scientists alike in an open and accessible collaborative environment. The contributed methods to WeFold include ICOS [3] for contact prediction (Bacardit Lab); CONCORD [4] for secondary structure prediction, BeST for ab initio beta-sheet topology prediction [5], contact prediction [6, 7], ICON for traveling-salesman problem based clustering (Floudas Lab); the online multiplayer game Foldit [8] and Rosetta [9] for sampling and selection (Baker Lab); UNRES [10-14] for sampling via Multiplexed Replica Exchange Molecular Dynamics (Liwo Lab and Scheraga Lab); GOAP [15] for knowledge-based scoring, TASSER [16] for sampling, SPICKER [17] for clustering (Skolnick Lab); KoBaMIN [18, 19] for refinement (Levitt Lab); APOLLO [20] for consensus-based quality assessment (Cheng Lab); Replica-Exchange Molecular Dynamics in GROMACS [21] for sampling (Univ. Sao Paulo, Brazil); and MESHI [22] for quality assessment (Keasar Lab). Performance of the synergistic branches will be compared with the performance of the base methods that comprise them. In total, the collaboration used over 1.5 million CPU hours and processed and evaluated over 8 million candidate structure models (~100x the size of the protein data bank [23]) in a 3.5 month period. All discussions and data generated during the collaboration are publically accessible at http://www.wefold.org. References: 1. Dill, K.A. and J.L. MacCallum, The Protein-Folding Problem, 50 Years On. Science, 2012. 338(6110): p. 1042-1046. 2. Soding, J., A. Biegert, and A.N. Lupas, The HHpred interactive server for protein homology detection and structure prediction. Nucleic Acids Research, 2005. 33(Web Server issue): p. W244-8. 3. Bacardit, J., et al., Contact map prediction using a large-scale ensemble of rule sets and the fusion of multiple predicted structural features.Bioinformatics, 2012. 4. Wei, Y., J. Thompson, and C.A. Floudas, CONCORD: a consensus method for protein secondary structure prediction via mixed integer linear optimization. Proceedings of the Royal Society A: Mathematical, Physical and Engineering Science, 2012. 468(2139): p. 831-850. 5. Subramani, A. and C.A. Floudas, β-sheet Topology Prediction with High Precision and Recall for β and Mixed α/β Proteins. PLoS ONE, 2012. 7(3): p. e32461. 6. Rajgaria, R., S.R. McAllister, and C.A. Floudas, Towards accurate residue–residue hydrophobic contact prediction for α helical proteins via integer linear optimization. Proteins: Structure, Function, and Bioinformatics, 2009. 74(4): p. 929-947. 7. Rajgaria, R., Y. Wei, and C.A. Floudas, Contact prediction for beta and alpha-beta proteins using integer linear optimization and its impact on the first principles 3D structure prediction method ASTRO-FOLD. Proteins, 2010. 78(8): p. 1825-46. 8. Cooper, S., et al., Predicting protein structures with a multiplayer online game. Nature, 2010. 466(7307): p. 756-760. 9. Leaver-Fay, A., et al., ROSETTA3: an object-oriented software suite for the simulation and design of macromolecules. Methods in enzymology, 2011. 487: p. 545-74. 10. Czaplewski, C., et al., Application of Multiplexed Replica Exchange Molecular Dynamics to the UNRES Force Field: Tests with α and α+β Proteins. Journal of Chemical Theory and Computation, 2009. 5(3): p. 627-640. 11. He, Y., et al., Exploring the parameter space of the coarse-grained UNRES force field by random search: Selecting a transferable medium-resolution force field. Journal of Computational Chemistry, 2009. 30(13): p. 2127-2135. 12. Liwo, A., et al., Simulation of protein structure and dynamics with the coarse-grained UNRES force field. Coarse-Graining of Condensed Phase and Biomolecular Systems, 2008. 1: p. 1391-1411. 13. Liwo, A., et al., Modification and Optimization of the United-Residue (UNRES) Potential Energy Function for Canonical…
Motivation The Protein Data Bank (PDB), the ultimate source for data in structural biology, is in... more Motivation The Protein Data Bank (PDB), the ultimate source for data in structural biology, is inherently imbalanced. To alleviate biases, virtually all structural biology studies use non-redundant subsets of the PDB, which include only a fraction of the available data. An alternative approach, dubbed redundancy-weighting, down-weights redundant entries rather than discarding them. This approach may be particularly helpful for Machine Learning (ML) methods that use the PDB as their source for data. Methods for Secondary Structure Prediction (SSP) have greatly improved over the years with recent studies achieving above 70% accuracy for 8-class (DSSP) prediction. As these methods typically incorporate machine learning techniques, training on redundancy-weighted datasets might improve accuracy, as well as pave the way toward larger and more informative secondary structure alphabets. Results This article compares the SSP performances of Deep Learning (DL) models trained on either redund...
Methods for reliably estimating the quality of 3D models of proteins are essential drivers for th... more Methods for reliably estimating the quality of 3D models of proteins are essential drivers for the wide adoption and serious acceptance of protein structure predictions by life scientists. In this paper, the most successful groups in CASP12 describe their latest methods for Estimates of Model Accuracy (EMA). We show that pure single model accuracy estimation methods has shown clear progress since CASP11; the three top methods (MESHI, ProQ3, SVMQA) all perform better than the top method of CASP11 (ProQ2). The pure single model accuracy estimation methods outperform quasi-single (ModFOLD6 variations) and consensus methods (Pcons, ModFOLDclust2, Pcomb-domain and Wallner) in model selection, but are still not as good as those methods in absolute model quality estimation and predictions of local quality. Finally, we show that when using contact based model quality measures (CAD, lDDT) the single model quality methods perform relatively better.
IEEE/ACM Transactions on Computational Biology and Bioinformatics, 2016
The function of a protein is determined by its structure, which creates a need for efficient meth... more The function of a protein is determined by its structure, which creates a need for efficient methods of protein structure determination to advance scientific and medical research. Because current experimental structure determination methods carry a high price tag, computational predictions are highly desirable. Given a protein sequence, computational methods produce numerous 3D structures known as decoys. However, selection of the best quality decoys is challenging as the end users can handle only a few ones. Therefore, scoring functions are central to decoy selection. They combine measurable features into a single number indicator of decoy quality. Unfortunately, current scoring functions do not consistently select the best decoys. Machine learning techniques offer great potential to improve decoy scoring. This paper presents two machine-learning based scoring functions to predict the quality of proteins structures, i.e., the similarity between the predicted structure and the experimental one without knowing the latter. We use different metrics to compare these scoring functions against three state-of-the-art scores. This is a first attempt at comparing different scoring functions using the same non-redundant dataset for training and testing and the same features. The results show that adding informative features may be more significant than the method used.
Background: Peptides have ubiquitous roles in all biological systems and are thus of interest in ... more Background: Peptides have ubiquitous roles in all biological systems and are thus of interest in both basic and applied research. The rational design of bioactive peptides could be greatly enhanced by an efficient method for accurately predicting the conformations that these molecules can adopt in solution. As a design process inevitably requires testing numerous molecules, an efficient method would require the calculations to be performed in vacuum. Results: Attempts to predict the conformations of cyclic peptides using a simulated annealing protocol with the Amber/OPLS potential in vacuum resulted, not unexpectedly, in overly packed, non-native conformations. We therefore empirically modified the potential by several cycles of structure prediction and function refinement until a good fit between experimental and predicted conformations was obtained. Three major modifications to the potential were required in order to reproduce the solution structures of cyclic peptides: explicit torsional energies for the peptide backbone torsional angles; explicit hydrogen-bonding energies for backbone hydrogen bonds; and a penalty for close approaches between uncharged and charged atoms. Conclusions: Using the modified potential, we predicted the solution conformations of cyclic peptides in the size range of 5-10 residues with reasonable accuracy.
Simulations of molecular systems typically handle interactions within non-bonded pairs. Generatin... more Simulations of molecular systems typically handle interactions within non-bonded pairs. Generating and updating a list of these pairs can be the most time-consuming part of energy calculations for large systems. Thus, efficient non-bonded list processing can speed up the energy calculations significantly. While the asymptotic complexity of current algorithms (namely O(N), where N is the number of particles) is probably the lowest possible, a wide space for optimization is still left. This article offers a heuristic extension to the previously suggested grid based algorithms. We show that, when the average particle movements are slow, simulation time can be reduced considerably. The proposed algorithm has been implemented in the DistanceMatrix class of the molecular modeling package MESHI. MESHI is freely available at <www.cs.bgu.ac.il/∼meshi>.
Every two years groups worldwide participate in the Critical Assessment of Protein Structure Pred... more Every two years groups worldwide participate in the Critical Assessment of Protein Structure Prediction (CASP) experiment to blindly test the strengths and weaknesses of their computational methods. CASP has significantly advanced the field but many hurdles still remain, which may require new ideas and collaborations. In 2012 a web-based effort called WeFold, was initiated to promote collaboration within the CASP community and attract researchers from other fields to contribute new ideas to CASP. Members of the WeFold coopetition (cooperation and competition) participated in CASP as individual teams, but also shared components of their methods to create hybrid pipelines and actively contributed to this effort. We assert that the scale and diversity of integrative prediction pipelines could not have been achieved by any individual lab or even by any collaboration among a few partners. The models contributed by the participating groups and generated by the pipelines are publicly avail...
Protein surfaces serve as an interface with the molecular environment and are thus tightly bound ... more Protein surfaces serve as an interface with the molecular environment and are thus tightly bound to proteins' function. On the surface, geometric and chemical complementarity to other molecules provides interaction specificity for ligand binding, docking of bio-macromolecules, and enzymatic catalysis.
ABSTRACT The protein structure prediction problem continues to elude scientists. Even though many... more ABSTRACT The protein structure prediction problem continues to elude scientists. Even though many new methods have been introduced, certain classes of prediction targets such as free modeling targets remain a challenge based on blind predictions in the several previous Critical Assessment of Techniques for Protein Structure Prediction (CASP) experiments [1]. To meet this challenge, a large-scale collaborative effort called WeFold was undertaken by thirteen labs, each with their own specialties and approaches in addressing the problem. In this talk, we will present the different methods or branches collaboratively designed and tested during the WeFold experiment, as well as their predictive ability, outcomes, and lessons learned. Independent branches involved in the collaborative effort yielded several high-ranking predictions among all group and method submissions in CASP10 for human, free modeling (template free), and refinement targets. Remarkably, two WeFold methods were able to produce the very best predictions in the refinement category in two different targets from a starting model for several different accuracy metrics. Contributions were made by junior and seasoned scientists alike in an open and accessible collaborative environment. The contributed methods to WeFold include ICOS [3] for contact prediction (Bacardit Lab); CONCORD [4] for secondary structure prediction, BeST for ab initio beta-sheet topology prediction [5], contact prediction [6, 7], ICON for traveling-salesman problem based clustering (Floudas Lab); the online multiplayer game Foldit [8] and Rosetta [9] for sampling and selection (Baker Lab); UNRES [10-14] for sampling via Multiplexed Replica Exchange Molecular Dynamics (Liwo Lab and Scheraga Lab); GOAP [15] for knowledge-based scoring, TASSER [16] for sampling, SPICKER [17] for clustering (Skolnick Lab); KoBaMIN [18, 19] for refinement (Levitt Lab); APOLLO [20] for consensus-based quality assessment (Cheng Lab); Replica-Exchange Molecular Dynamics in GROMACS [21] for sampling (Univ. Sao Paulo, Brazil); and MESHI [22] for quality assessment (Keasar Lab). Performance of the synergistic branches will be compared with the performance of the base methods that comprise them. In total, the collaboration used over 1.5 million CPU hours and processed and evaluated over 8 million candidate structure models (~100x the size of the protein data bank [23]) in a 3.5 month period. All discussions and data generated during the collaboration are publically accessible at http://www.wefold.org. References: 1. Dill, K.A. and J.L. MacCallum, The Protein-Folding Problem, 50 Years On. Science, 2012. 338(6110): p. 1042-1046. 2. Soding, J., A. Biegert, and A.N. Lupas, The HHpred interactive server for protein homology detection and structure prediction. Nucleic Acids Research, 2005. 33(Web Server issue): p. W244-8. 3. Bacardit, J., et al., Contact map prediction using a large-scale ensemble of rule sets and the fusion of multiple predicted structural features.Bioinformatics, 2012. 4. Wei, Y., J. Thompson, and C.A. Floudas, CONCORD: a consensus method for protein secondary structure prediction via mixed integer linear optimization. Proceedings of the Royal Society A: Mathematical, Physical and Engineering Science, 2012. 468(2139): p. 831-850. 5. Subramani, A. and C.A. Floudas, β-sheet Topology Prediction with High Precision and Recall for β and Mixed α/β Proteins. PLoS ONE, 2012. 7(3): p. e32461. 6. Rajgaria, R., S.R. McAllister, and C.A. Floudas, Towards accurate residue–residue hydrophobic contact prediction for α helical proteins via integer linear optimization. Proteins: Structure, Function, and Bioinformatics, 2009. 74(4): p. 929-947. 7. Rajgaria, R., Y. Wei, and C.A. Floudas, Contact prediction for beta and alpha-beta proteins using integer linear optimization and its impact on the first principles 3D structure prediction method ASTRO-FOLD. Proteins, 2010. 78(8): p. 1825-46. 8. Cooper, S., et al., Predicting protein structures with a multiplayer online game. Nature, 2010. 466(7307): p. 756-760. 9. Leaver-Fay, A., et al., ROSETTA3: an object-oriented software suite for the simulation and design of macromolecules. Methods in enzymology, 2011. 487: p. 545-74. 10. Czaplewski, C., et al., Application of Multiplexed Replica Exchange Molecular Dynamics to the UNRES Force Field: Tests with α and α+β Proteins. Journal of Chemical Theory and Computation, 2009. 5(3): p. 627-640. 11. He, Y., et al., Exploring the parameter space of the coarse-grained UNRES force field by random search: Selecting a transferable medium-resolution force field. Journal of Computational Chemistry, 2009. 30(13): p. 2127-2135. 12. Liwo, A., et al., Simulation of protein structure and dynamics with the coarse-grained UNRES force field. Coarse-Graining of Condensed Phase and Biomolecular Systems, 2008. 1: p. 1391-1411. 13. Liwo, A., et al., Modification and Optimization of the United-Residue (UNRES) Potential Energy Function for Canonical…
All eukaryotic mRNAs possess a 5-cap (m 7 GpppN) that is recognized by a family of cap-binding pr... more All eukaryotic mRNAs possess a 5-cap (m 7 GpppN) that is recognized by a family of cap-binding proteins. These participate in various processes, such as RNA transport and stabilization, as well as in assembly of the translation initiation complex. The 5-cap of trypanosomatids is complex; in addition to 7-methyl guanosine, it includes unique modifications on the first four transcribed nucleotides, and is thus denoted cap-4. Here we analyze a cap-binding protein of Leishmania, in an attempt to understand the structural features that promote its binding to this unusual cap. LeishIF4E-1, a homolog of eIF4E, contains the conserved cap-binding pocket, similar to its mouse counterpart. The mouse eIF4E has a higher K as for all cap analogs tested, as compared with LeishIF4E-1. However, whereas the mouse eIF4E shows a fivefold higher affinity for m 7 GTP than for a chemically synthesized cap-4 structure, LeishIF4E-1 shows similar affinities for both ligands. A sequence alignment shows that LeishIF4E-1 lacks the region that parallels the C terminus in the murine eIF4E. Truncation of this region in the mouse protein reduces the difference that is observed between its binding to m 7 GTP and cap-4, prior to this deletion. We hypothesize that variations in the structure of LeishIF4E-1, possibly also the absence of a region that is homologous to the C terminus of the mouse protein, promote its ability to interact with the cap-4 structure. LeishIF4E-1 is distributed in the cytoplasm, but its function is not clear yet, because it cannot substitute the mammalian eIF4E in a rabbit reticulocyte in vitro translation system. ; fax: 972-8-6479185.
Proteins: Structure, Function, and Bioinformatics, 2014
The protein structure prediction problem continues to elude scientists. Despite the introduction ... more The protein structure prediction problem continues to elude scientists. Despite the introduction of many methods, only modest gains were made over the last decade for certain classes of prediction targets. To address this challenge, a socialmedia based worldwide collaborative effort, named WeFold, was undertaken by 13 labs. During the collaboration, the laboratories were simultaneously competing with each other. Here, we present the first attempt at "coopetition" in scientific research applied to the protein structure prediction and refinement problems. The coopetition was possible by allowing the participating labs to contribute different components of their protein structure prediction pipelines and create new hybrid pipelines that they tested during CASP10. This manuscript describes both successes and areas needing improvement as identified throughout the first WeFold experiment and discusses the efforts that are underway to advance this initiative. A footprint of all contributions and structures are publicly accessible at http://www.wefold.org.
The structural reorganization of the actin cytoskeleton is facilitated through the action of moto... more The structural reorganization of the actin cytoskeleton is facilitated through the action of motor proteins that crosslink the actin filaments and transport them relative to each other. Here, we present a combined experimental-computational study that probes the dynamic evolution of mixtures of actin filaments and clusters of myosin motors. While on small spatial and temporal scales the system behaves in a very noisy manner, on larger scales it evolves into several well distinct patterns such as bundles, asters and networks. These patterns are characterized by junctions with high connectivity, whose formation is possible due to the organization of the motors in &amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;#39;oligoclusters&amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;#39; (intermediate-size aggregates). The simulations reveal that the self-organization process proceeds through a series of hierarchical steps, starting from local microscopic moves and ranging up to the macroscopic large scales where the steady-state structures are formed. Our results shed light on the mechanisms involved in processes such as cytokinesis and cellular contractility, where myosin motors organized in clusters operate cooperatively to induce the structural organization of cytoskeletal networks.
Transfer of the green algae Chlamydomonas reinhardtii from low light to high light generated an o... more Transfer of the green algae Chlamydomonas reinhardtii from low light to high light generated an oxidative stress that led to a dramatic arrest in the synthesis of the large subunit (LSU) of ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco). The translational arrest correlated with transient changes in the intracellular levels of reactive oxygen species and with shifting the glutathione pool toward its oxidized form (Irihimovitch, V., and Shapira, M. (2000) J. Biol. Chem. 275, 16289-16295). Here we examined how the redox potential of glutathione affected the RNA-protein interactions with the 5&amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;#39;-untranslated region of rbcL. This RNA region specifically binds a group of proteins with molecular masses of 81, 62, 51, and 47 kDa in UV-cross-linking experiments under reducing conditions. Binding of these proteins was interrupted by exposure to oxidizing conditions (GSSG), and a new protein of 55 kDa was shown to interact with the RNA. The 55-kDa protein comigrated with Rubisco LSU in one- and two-dimensional gels, and its RNA binding activity was further verified by using the purified protein in UV-cross-linking experiments under oxidizing conditions. However, the LSU of purified and oxidized Rubisco bound to RNA in a sequence-independent manner. A remarkable structural similarity was found between the amino-terminal domain of Rubisco LSU in C. reinhardtii and the RNA binding domain, a highly prevailing motif among RNA-binding proteins. It appears from the crystal structure of Rubisco that the amino terminus of LSU is buried within the holoenzyme. We propose that under oxidizing conditions it is exposed to the surface and can, therefore, bind RNA. Accordingly, a recombinant form of the polypeptide domain that corresponds to the amino terminus of LSU was found to bind RNA in vitro with or without GSSG.
ABSTRACT The protein structure prediction problem continues to elude scientists. Even though many... more ABSTRACT The protein structure prediction problem continues to elude scientists. Even though many new methods have been introduced, certain classes of prediction targets such as free modeling targets remain a challenge based on blind predictions in the several previous Critical Assessment of Techniques for Protein Structure Prediction (CASP) experiments [1]. To meet this challenge, a large-scale collaborative effort called WeFold was undertaken by thirteen labs, each with their own specialties and approaches in addressing the problem. In this talk, we will present the different methods or branches collaboratively designed and tested during the WeFold experiment, as well as their predictive ability, outcomes, and lessons learned. Independent branches involved in the collaborative effort yielded several high-ranking predictions among all group and method submissions in CASP10 for human, free modeling (template free), and refinement targets. Remarkably, two WeFold methods were able to produce the very best predictions in the refinement category in two different targets from a starting model for several different accuracy metrics. Contributions were made by junior and seasoned scientists alike in an open and accessible collaborative environment. The contributed methods to WeFold include ICOS [3] for contact prediction (Bacardit Lab); CONCORD [4] for secondary structure prediction, BeST for ab initio beta-sheet topology prediction [5], contact prediction [6, 7], ICON for traveling-salesman problem based clustering (Floudas Lab); the online multiplayer game Foldit [8] and Rosetta [9] for sampling and selection (Baker Lab); UNRES [10-14] for sampling via Multiplexed Replica Exchange Molecular Dynamics (Liwo Lab and Scheraga Lab); GOAP [15] for knowledge-based scoring, TASSER [16] for sampling, SPICKER [17] for clustering (Skolnick Lab); KoBaMIN [18, 19] for refinement (Levitt Lab); APOLLO [20] for consensus-based quality assessment (Cheng Lab); Replica-Exchange Molecular Dynamics in GROMACS [21] for sampling (Univ. Sao Paulo, Brazil); and MESHI [22] for quality assessment (Keasar Lab). Performance of the synergistic branches will be compared with the performance of the base methods that comprise them. In total, the collaboration used over 1.5 million CPU hours and processed and evaluated over 8 million candidate structure models (~100x the size of the protein data bank [23]) in a 3.5 month period. All discussions and data generated during the collaboration are publically accessible at http://www.wefold.org. References: 1. Dill, K.A. and J.L. MacCallum, The Protein-Folding Problem, 50 Years On. Science, 2012. 338(6110): p. 1042-1046. 2. Soding, J., A. Biegert, and A.N. Lupas, The HHpred interactive server for protein homology detection and structure prediction. Nucleic Acids Research, 2005. 33(Web Server issue): p. W244-8. 3. Bacardit, J., et al., Contact map prediction using a large-scale ensemble of rule sets and the fusion of multiple predicted structural features.Bioinformatics, 2012. 4. Wei, Y., J. Thompson, and C.A. Floudas, CONCORD: a consensus method for protein secondary structure prediction via mixed integer linear optimization. Proceedings of the Royal Society A: Mathematical, Physical and Engineering Science, 2012. 468(2139): p. 831-850. 5. Subramani, A. and C.A. Floudas, β-sheet Topology Prediction with High Precision and Recall for β and Mixed α/β Proteins. PLoS ONE, 2012. 7(3): p. e32461. 6. Rajgaria, R., S.R. McAllister, and C.A. Floudas, Towards accurate residue–residue hydrophobic contact prediction for α helical proteins via integer linear optimization. Proteins: Structure, Function, and Bioinformatics, 2009. 74(4): p. 929-947. 7. Rajgaria, R., Y. Wei, and C.A. Floudas, Contact prediction for beta and alpha-beta proteins using integer linear optimization and its impact on the first principles 3D structure prediction method ASTRO-FOLD. Proteins, 2010. 78(8): p. 1825-46. 8. Cooper, S., et al., Predicting protein structures with a multiplayer online game. Nature, 2010. 466(7307): p. 756-760. 9. Leaver-Fay, A., et al., ROSETTA3: an object-oriented software suite for the simulation and design of macromolecules. Methods in enzymology, 2011. 487: p. 545-74. 10. Czaplewski, C., et al., Application of Multiplexed Replica Exchange Molecular Dynamics to the UNRES Force Field: Tests with α and α+β Proteins. Journal of Chemical Theory and Computation, 2009. 5(3): p. 627-640. 11. He, Y., et al., Exploring the parameter space of the coarse-grained UNRES force field by random search: Selecting a transferable medium-resolution force field. Journal of Computational Chemistry, 2009. 30(13): p. 2127-2135. 12. Liwo, A., et al., Simulation of protein structure and dynamics with the coarse-grained UNRES force field. Coarse-Graining of Condensed Phase and Biomolecular Systems, 2008. 1: p. 1391-1411. 13. Liwo, A., et al., Modification and Optimization of the United-Residue (UNRES) Potential Energy Function for Canonical…
Motivation The Protein Data Bank (PDB), the ultimate source for data in structural biology, is in... more Motivation The Protein Data Bank (PDB), the ultimate source for data in structural biology, is inherently imbalanced. To alleviate biases, virtually all structural biology studies use non-redundant subsets of the PDB, which include only a fraction of the available data. An alternative approach, dubbed redundancy-weighting, down-weights redundant entries rather than discarding them. This approach may be particularly helpful for Machine Learning (ML) methods that use the PDB as their source for data. Methods for Secondary Structure Prediction (SSP) have greatly improved over the years with recent studies achieving above 70% accuracy for 8-class (DSSP) prediction. As these methods typically incorporate machine learning techniques, training on redundancy-weighted datasets might improve accuracy, as well as pave the way toward larger and more informative secondary structure alphabets. Results This article compares the SSP performances of Deep Learning (DL) models trained on either redund...
Methods for reliably estimating the quality of 3D models of proteins are essential drivers for th... more Methods for reliably estimating the quality of 3D models of proteins are essential drivers for the wide adoption and serious acceptance of protein structure predictions by life scientists. In this paper, the most successful groups in CASP12 describe their latest methods for Estimates of Model Accuracy (EMA). We show that pure single model accuracy estimation methods has shown clear progress since CASP11; the three top methods (MESHI, ProQ3, SVMQA) all perform better than the top method of CASP11 (ProQ2). The pure single model accuracy estimation methods outperform quasi-single (ModFOLD6 variations) and consensus methods (Pcons, ModFOLDclust2, Pcomb-domain and Wallner) in model selection, but are still not as good as those methods in absolute model quality estimation and predictions of local quality. Finally, we show that when using contact based model quality measures (CAD, lDDT) the single model quality methods perform relatively better.
IEEE/ACM Transactions on Computational Biology and Bioinformatics, 2016
The function of a protein is determined by its structure, which creates a need for efficient meth... more The function of a protein is determined by its structure, which creates a need for efficient methods of protein structure determination to advance scientific and medical research. Because current experimental structure determination methods carry a high price tag, computational predictions are highly desirable. Given a protein sequence, computational methods produce numerous 3D structures known as decoys. However, selection of the best quality decoys is challenging as the end users can handle only a few ones. Therefore, scoring functions are central to decoy selection. They combine measurable features into a single number indicator of decoy quality. Unfortunately, current scoring functions do not consistently select the best decoys. Machine learning techniques offer great potential to improve decoy scoring. This paper presents two machine-learning based scoring functions to predict the quality of proteins structures, i.e., the similarity between the predicted structure and the experimental one without knowing the latter. We use different metrics to compare these scoring functions against three state-of-the-art scores. This is a first attempt at comparing different scoring functions using the same non-redundant dataset for training and testing and the same features. The results show that adding informative features may be more significant than the method used.
Background: Peptides have ubiquitous roles in all biological systems and are thus of interest in ... more Background: Peptides have ubiquitous roles in all biological systems and are thus of interest in both basic and applied research. The rational design of bioactive peptides could be greatly enhanced by an efficient method for accurately predicting the conformations that these molecules can adopt in solution. As a design process inevitably requires testing numerous molecules, an efficient method would require the calculations to be performed in vacuum. Results: Attempts to predict the conformations of cyclic peptides using a simulated annealing protocol with the Amber/OPLS potential in vacuum resulted, not unexpectedly, in overly packed, non-native conformations. We therefore empirically modified the potential by several cycles of structure prediction and function refinement until a good fit between experimental and predicted conformations was obtained. Three major modifications to the potential were required in order to reproduce the solution structures of cyclic peptides: explicit torsional energies for the peptide backbone torsional angles; explicit hydrogen-bonding energies for backbone hydrogen bonds; and a penalty for close approaches between uncharged and charged atoms. Conclusions: Using the modified potential, we predicted the solution conformations of cyclic peptides in the size range of 5-10 residues with reasonable accuracy.
Simulations of molecular systems typically handle interactions within non-bonded pairs. Generatin... more Simulations of molecular systems typically handle interactions within non-bonded pairs. Generating and updating a list of these pairs can be the most time-consuming part of energy calculations for large systems. Thus, efficient non-bonded list processing can speed up the energy calculations significantly. While the asymptotic complexity of current algorithms (namely O(N), where N is the number of particles) is probably the lowest possible, a wide space for optimization is still left. This article offers a heuristic extension to the previously suggested grid based algorithms. We show that, when the average particle movements are slow, simulation time can be reduced considerably. The proposed algorithm has been implemented in the DistanceMatrix class of the molecular modeling package MESHI. MESHI is freely available at <www.cs.bgu.ac.il/∼meshi>.
Every two years groups worldwide participate in the Critical Assessment of Protein Structure Pred... more Every two years groups worldwide participate in the Critical Assessment of Protein Structure Prediction (CASP) experiment to blindly test the strengths and weaknesses of their computational methods. CASP has significantly advanced the field but many hurdles still remain, which may require new ideas and collaborations. In 2012 a web-based effort called WeFold, was initiated to promote collaboration within the CASP community and attract researchers from other fields to contribute new ideas to CASP. Members of the WeFold coopetition (cooperation and competition) participated in CASP as individual teams, but also shared components of their methods to create hybrid pipelines and actively contributed to this effort. We assert that the scale and diversity of integrative prediction pipelines could not have been achieved by any individual lab or even by any collaboration among a few partners. The models contributed by the participating groups and generated by the pipelines are publicly avail...
Protein surfaces serve as an interface with the molecular environment and are thus tightly bound ... more Protein surfaces serve as an interface with the molecular environment and are thus tightly bound to proteins' function. On the surface, geometric and chemical complementarity to other molecules provides interaction specificity for ligand binding, docking of bio-macromolecules, and enzymatic catalysis.
ABSTRACT The protein structure prediction problem continues to elude scientists. Even though many... more ABSTRACT The protein structure prediction problem continues to elude scientists. Even though many new methods have been introduced, certain classes of prediction targets such as free modeling targets remain a challenge based on blind predictions in the several previous Critical Assessment of Techniques for Protein Structure Prediction (CASP) experiments [1]. To meet this challenge, a large-scale collaborative effort called WeFold was undertaken by thirteen labs, each with their own specialties and approaches in addressing the problem. In this talk, we will present the different methods or branches collaboratively designed and tested during the WeFold experiment, as well as their predictive ability, outcomes, and lessons learned. Independent branches involved in the collaborative effort yielded several high-ranking predictions among all group and method submissions in CASP10 for human, free modeling (template free), and refinement targets. Remarkably, two WeFold methods were able to produce the very best predictions in the refinement category in two different targets from a starting model for several different accuracy metrics. Contributions were made by junior and seasoned scientists alike in an open and accessible collaborative environment. The contributed methods to WeFold include ICOS [3] for contact prediction (Bacardit Lab); CONCORD [4] for secondary structure prediction, BeST for ab initio beta-sheet topology prediction [5], contact prediction [6, 7], ICON for traveling-salesman problem based clustering (Floudas Lab); the online multiplayer game Foldit [8] and Rosetta [9] for sampling and selection (Baker Lab); UNRES [10-14] for sampling via Multiplexed Replica Exchange Molecular Dynamics (Liwo Lab and Scheraga Lab); GOAP [15] for knowledge-based scoring, TASSER [16] for sampling, SPICKER [17] for clustering (Skolnick Lab); KoBaMIN [18, 19] for refinement (Levitt Lab); APOLLO [20] for consensus-based quality assessment (Cheng Lab); Replica-Exchange Molecular Dynamics in GROMACS [21] for sampling (Univ. Sao Paulo, Brazil); and MESHI [22] for quality assessment (Keasar Lab). Performance of the synergistic branches will be compared with the performance of the base methods that comprise them. In total, the collaboration used over 1.5 million CPU hours and processed and evaluated over 8 million candidate structure models (~100x the size of the protein data bank [23]) in a 3.5 month period. All discussions and data generated during the collaboration are publically accessible at http://www.wefold.org. References: 1. Dill, K.A. and J.L. MacCallum, The Protein-Folding Problem, 50 Years On. Science, 2012. 338(6110): p. 1042-1046. 2. Soding, J., A. Biegert, and A.N. Lupas, The HHpred interactive server for protein homology detection and structure prediction. Nucleic Acids Research, 2005. 33(Web Server issue): p. W244-8. 3. Bacardit, J., et al., Contact map prediction using a large-scale ensemble of rule sets and the fusion of multiple predicted structural features.Bioinformatics, 2012. 4. Wei, Y., J. Thompson, and C.A. Floudas, CONCORD: a consensus method for protein secondary structure prediction via mixed integer linear optimization. Proceedings of the Royal Society A: Mathematical, Physical and Engineering Science, 2012. 468(2139): p. 831-850. 5. Subramani, A. and C.A. Floudas, β-sheet Topology Prediction with High Precision and Recall for β and Mixed α/β Proteins. PLoS ONE, 2012. 7(3): p. e32461. 6. Rajgaria, R., S.R. McAllister, and C.A. Floudas, Towards accurate residue–residue hydrophobic contact prediction for α helical proteins via integer linear optimization. Proteins: Structure, Function, and Bioinformatics, 2009. 74(4): p. 929-947. 7. Rajgaria, R., Y. Wei, and C.A. Floudas, Contact prediction for beta and alpha-beta proteins using integer linear optimization and its impact on the first principles 3D structure prediction method ASTRO-FOLD. Proteins, 2010. 78(8): p. 1825-46. 8. Cooper, S., et al., Predicting protein structures with a multiplayer online game. Nature, 2010. 466(7307): p. 756-760. 9. Leaver-Fay, A., et al., ROSETTA3: an object-oriented software suite for the simulation and design of macromolecules. Methods in enzymology, 2011. 487: p. 545-74. 10. Czaplewski, C., et al., Application of Multiplexed Replica Exchange Molecular Dynamics to the UNRES Force Field: Tests with α and α+β Proteins. Journal of Chemical Theory and Computation, 2009. 5(3): p. 627-640. 11. He, Y., et al., Exploring the parameter space of the coarse-grained UNRES force field by random search: Selecting a transferable medium-resolution force field. Journal of Computational Chemistry, 2009. 30(13): p. 2127-2135. 12. Liwo, A., et al., Simulation of protein structure and dynamics with the coarse-grained UNRES force field. Coarse-Graining of Condensed Phase and Biomolecular Systems, 2008. 1: p. 1391-1411. 13. Liwo, A., et al., Modification and Optimization of the United-Residue (UNRES) Potential Energy Function for Canonical…
All eukaryotic mRNAs possess a 5-cap (m 7 GpppN) that is recognized by a family of cap-binding pr... more All eukaryotic mRNAs possess a 5-cap (m 7 GpppN) that is recognized by a family of cap-binding proteins. These participate in various processes, such as RNA transport and stabilization, as well as in assembly of the translation initiation complex. The 5-cap of trypanosomatids is complex; in addition to 7-methyl guanosine, it includes unique modifications on the first four transcribed nucleotides, and is thus denoted cap-4. Here we analyze a cap-binding protein of Leishmania, in an attempt to understand the structural features that promote its binding to this unusual cap. LeishIF4E-1, a homolog of eIF4E, contains the conserved cap-binding pocket, similar to its mouse counterpart. The mouse eIF4E has a higher K as for all cap analogs tested, as compared with LeishIF4E-1. However, whereas the mouse eIF4E shows a fivefold higher affinity for m 7 GTP than for a chemically synthesized cap-4 structure, LeishIF4E-1 shows similar affinities for both ligands. A sequence alignment shows that LeishIF4E-1 lacks the region that parallels the C terminus in the murine eIF4E. Truncation of this region in the mouse protein reduces the difference that is observed between its binding to m 7 GTP and cap-4, prior to this deletion. We hypothesize that variations in the structure of LeishIF4E-1, possibly also the absence of a region that is homologous to the C terminus of the mouse protein, promote its ability to interact with the cap-4 structure. LeishIF4E-1 is distributed in the cytoplasm, but its function is not clear yet, because it cannot substitute the mammalian eIF4E in a rabbit reticulocyte in vitro translation system. ; fax: 972-8-6479185.
Proteins: Structure, Function, and Bioinformatics, 2014
The protein structure prediction problem continues to elude scientists. Despite the introduction ... more The protein structure prediction problem continues to elude scientists. Despite the introduction of many methods, only modest gains were made over the last decade for certain classes of prediction targets. To address this challenge, a socialmedia based worldwide collaborative effort, named WeFold, was undertaken by 13 labs. During the collaboration, the laboratories were simultaneously competing with each other. Here, we present the first attempt at "coopetition" in scientific research applied to the protein structure prediction and refinement problems. The coopetition was possible by allowing the participating labs to contribute different components of their protein structure prediction pipelines and create new hybrid pipelines that they tested during CASP10. This manuscript describes both successes and areas needing improvement as identified throughout the first WeFold experiment and discusses the efforts that are underway to advance this initiative. A footprint of all contributions and structures are publicly accessible at http://www.wefold.org.
The structural reorganization of the actin cytoskeleton is facilitated through the action of moto... more The structural reorganization of the actin cytoskeleton is facilitated through the action of motor proteins that crosslink the actin filaments and transport them relative to each other. Here, we present a combined experimental-computational study that probes the dynamic evolution of mixtures of actin filaments and clusters of myosin motors. While on small spatial and temporal scales the system behaves in a very noisy manner, on larger scales it evolves into several well distinct patterns such as bundles, asters and networks. These patterns are characterized by junctions with high connectivity, whose formation is possible due to the organization of the motors in &amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;#39;oligoclusters&amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;#39; (intermediate-size aggregates). The simulations reveal that the self-organization process proceeds through a series of hierarchical steps, starting from local microscopic moves and ranging up to the macroscopic large scales where the steady-state structures are formed. Our results shed light on the mechanisms involved in processes such as cytokinesis and cellular contractility, where myosin motors organized in clusters operate cooperatively to induce the structural organization of cytoskeletal networks.
Transfer of the green algae Chlamydomonas reinhardtii from low light to high light generated an o... more Transfer of the green algae Chlamydomonas reinhardtii from low light to high light generated an oxidative stress that led to a dramatic arrest in the synthesis of the large subunit (LSU) of ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco). The translational arrest correlated with transient changes in the intracellular levels of reactive oxygen species and with shifting the glutathione pool toward its oxidized form (Irihimovitch, V., and Shapira, M. (2000) J. Biol. Chem. 275, 16289-16295). Here we examined how the redox potential of glutathione affected the RNA-protein interactions with the 5&amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;amp;#39;-untranslated region of rbcL. This RNA region specifically binds a group of proteins with molecular masses of 81, 62, 51, and 47 kDa in UV-cross-linking experiments under reducing conditions. Binding of these proteins was interrupted by exposure to oxidizing conditions (GSSG), and a new protein of 55 kDa was shown to interact with the RNA. The 55-kDa protein comigrated with Rubisco LSU in one- and two-dimensional gels, and its RNA binding activity was further verified by using the purified protein in UV-cross-linking experiments under oxidizing conditions. However, the LSU of purified and oxidized Rubisco bound to RNA in a sequence-independent manner. A remarkable structural similarity was found between the amino-terminal domain of Rubisco LSU in C. reinhardtii and the RNA binding domain, a highly prevailing motif among RNA-binding proteins. It appears from the crystal structure of Rubisco that the amino terminus of LSU is buried within the holoenzyme. We propose that under oxidizing conditions it is exposed to the surface and can, therefore, bind RNA. Accordingly, a recombinant form of the polypeptide domain that corresponds to the amino terminus of LSU was found to bind RNA in vitro with or without GSSG.
Uploads
Papers by Chen Keasar