Computational Sciences and Engineering

Ammonium transport is mediated by membrane proteins of the ubiquitous Amt/Rh family. Despite the availability of different X-ray structures that provide many insights on the ammonium permeation process, the molecular details of its mechanism remain controversial. The X-ray structures have revealed that the pore of the Amt and Rh proteins is characterized by a hydrophobic portion about 12 Å long in which electronic density was observed in crystallographic study of AmtB from Escherichia coli. This electronic density was initially only observed when crystals were grown in presence of ammonium salt and was thus attributed to ammonia (NH 3 ) molecules, and lead the authors to suggest that the conduction mechanism in the Amt/Rh proteins involves the single-file diffusion of NH 3 molecules. However, other X-ray crystallography results and molecular mechanics simulations suggest that the pore of AmtB could also be filled with water molecules. The possible presence of water molecules in the pore lumen calls for a reassessment of the growing consensus that Amt/Rh proteins work as plain NH 3 channels. Indeed, functional experiments on plant ammonium transporters and rhesus proteins suggest a variety of permeation mechanisms including the passive diffusion of NH 3 , the antiport of NH 4 + /H + , the transport of NH 4 + , or the cotransport of NH 3 /H + . We discuss these mechanisms in light of some recent functional and simulation studies on the AmtB transporter and illustrate how they can be reconciled with the available high resolution X-ray data. #

The folding of a protein is studied as it grows residue by residue from the N-terminus and enters an environment that stabilizes the folded state. This mode of folding of a growing chain is different from refolding where the full chain folds from a disordered initial configuration to the native state. We propose a sequential dynamic optimization method that computes the evolution of optimum folding pathways as amino acid residues are added to the peptide chain one by one. The dynamic optimization formulation is deterministic and uses Newton's equations of motion and a Go-type potential that establishes the native contacts and excluded volume effects. The method predicts the optimal energy-minimizing path among all the alternative feasible pathways. As two examples, the folding of the chicken villin headpiece, a 36-residue protein, and chymotrypsin inhibitor 2 (CI2), a 64-residue protein, are studied. Results on the villin headpiece show significant differences from the refolding of the same chain studied previously. Results on CI2 mostly agree with the results of refolding experiments and computational work.

The ammonium transporters of the Amt/Rh family facilitate the diffusion of ammonium across cellular membranes. Functional data suggest that Amt proteins, notably found in plants, transport the ammonium ion (NH 4 + ), whereas human Rhesus (Rh) proteins transport ammonia (NH 3 ). Comparison between the X-ray structures of the prokaryotic AmtB, assumed to be representative of Amt proteins, and the human RhCG reveals important differences at the level of their pore. Despite these important functional and structural differences between Amt and Rh proteins, studies of the AmtB transporter have led to the suggestion that proteins of both subfamilies work according to the same mechanism and transport ammonia. We performed molecular dynamics simulations of the AmtB and RhCG proteins under different water and ammonia occupancy states of their pore. Free energy calculations suggest that the probability of finding NH 3 molecules in the pore of AmtB is negligible in comparison to finding water. The presence of water in the pore of AmtB could support the transport of proton. The pore lumen of RhCG is found to be more hydrophobic due to the presence of a phenylalanine conserved among Rh proteins. Simulations of RhCG also reveal that the signature histidine dyad is occasionally exposed to the extracellular bulk, which is never observed in AmtB. These different hydration patterns are consistent with the idea that Amt and Rh proteins are not functionally equivalent and that permeation takes place according to two distinct mechanisms.

The nonlocal nature of the protein-ligand binding problem is investigated via the Gaussian Network Model with which the residues lying along interaction pathways in a protein and the residues at the binding site are predicted. The predictions of the binding site residues are verified by using several benchmark systems where the topology of the unbound protein and the bound protein-ligand complex are known. Predictions are made on the unbound protein. Agreement of results with the bound complexes indicates that the information for binding resides in the unbound protein. Cliques that consist of three or more residues that are far apart along the primary structure but are in contact in the folded structure are shown to be important determinants of the binding problem. Comparison with known structures shows that the predictive capability of the method is significant.

We present a new database of computational hot spots in protein interfaces: HotSprint. Hot spots are residues comprising only a small fraction of interfaces yet accounting for the majority of the binding energy. HotSprint contains data for 35 776 protein interfaces among 49 512 protein interfaces extracted from the multi-chain structures in Protein Data Bank (PDB) as of February 2006. The conserved residues in interfaces with certain buried accessible solvent area (ASA) and complex ASA thresholds are flagged as computational hot spots. The predicted hot spots are observed to correlate with the experimental hot spots with an accuracy of 76%. Several machine-learning methods (SVM, Decision Trees and Decision Lists) are also applied to predict hot spots, results reveal that our empirical approach performs better than the others. A web interface for the HotSprint database allows users to browse and query the hot spots in protein interfaces. HotSprint is available at http://prism.ccbb.ku.edu.tr/hotsprint; and it provides information for interface residues that are functionally and structurally important as well as the evolutionary history and solvent accessibility of residues in interfaces.

The energy distribution along the protein–protein interface is not homogenous; certain residues contribute more to the binding free energy, called ‘hot spots’. Here, we present a web server, HotPoint, which predicts hot spots in protein interfaces using an empirical model. The empirical model incorporates a few simple rules consisting of occlusion from solvent and total knowledge-based pair potentials of residues. The prediction model is computationally efficient and achieves high accuracy of 70%. The input to the HotPoint server is a protein complex and two chain identifiers that form an interface. The server provides the hot spot prediction results, a table of residue properties and an interactive 3D visualization of the complex with hot spots highlighted. Results are also downloadable as text files. This web server can be used for analysis of any protein–protein interface which can be utilized by researchers working on binding sites characterization and rational design of small molecules for protein interactions. HotPoint is accessible at http://prism.ccbb.ku.edu.tr/hotpoint.

In this work, we present a computational scheme for finding high probability conformations of peptides. The scheme calculates the probability of a given conformation of the given peptide sequence using the probability distribution of torsion states. Dependence of the states of a residue on the states of its first neighbors along the chain is considered. Prior probabilities of torsion states are obtained from a coil library. Posterior probabilities are calculated by the matrix multiplication Rotational Isomeric States Model of polymer theory. The conformation of a peptide with highest probability is determined by using a hidden Markov model Viterbi algorithm. First, the probability distribution of the torsion states of the residues is obtained. Using the highest probability torsion state, one can generate, step by step, states with lower probabilities. To validate the method, the highest probability state of residues in a given sequence is calculated and compared with probabilities obtained from the Coil Databank. Predictions based on the method are 32% better than predictions based on the most probable states of residues. The ensemble of ''n'' high probability conformations of a given protein is also determined using the Viterbi algorithm with multistep backtracking.

Word sense induction aims to discover different senses of a word from a corpus by using unsupervised learning approaches. Once a sense inventory is obtained for an ambiguous word, word sense discrimination approaches choose the best-fitting single sense for a given context from the induced sense inventory. However, there may not be a clear distinction between one sense and another, although for a context, more than one induced sense can be suitable. Graded word sense method allows for labeling a word in more than one sense. In contrast to the most common approach which is to apply clustering or graph partitioning on a representation of first or second order co-occurrences of a word, we propose a system that creates a substitute vector for each target word from the most likely substitutes suggested by a statistical language model. Word samples are then taken according to probabilities of these substitutes and the results of the co-occurrence model are clustered. This approach outperforms the other systems on graded word sense induction task in SemEval-2013.

The success of a recommendation algorithm is typically measured by its ability to predict rating values of items. Although accuracy in rating value prediction is an important property of a recommendation algorithm there are other properties of recommendation algorithms which are important for user satisfaction. One such property is the diversity of recommendations. It has been recognized that being able to recommend a diverse set of items plays an important role in user satisfaction. One convenient approach for diversification is to use the rating patterns of items. However, in what sense the resulting lists will be diversified is not clear. In order to assess this we explore the relationship between rating similarity and content similarity of items. We discuss the experimental results and the possible implications of our findings.

In this paper, we describe our unsupervised method submitted to the Cross-Level Semantic Similarity task in Semeval 2014 that computes semantic similarity between two different sized text fragments. Our method models each text fragment by using the cooccurrence statistics of either occurred words or their substitutes. The co-occurrence modeling step provides dense, low-dimensional embedding for each fragment which allows us to calculate semantic similarity using various similarity metrics. Although our current model avoids the syntactic information, we achieved promising results and outperformed all baselines.

The Swi1 and Swi3 proteins are required for mat1 imprinting and mating-type switching in Schizosaccharomyces pombe, where they mediate a pause of leading-strand replication in response to a lagging-strand signal. In addition, Swi1 has been demonstrated to be involved in the checkpoint response to stalled replication forks, as was described for the Saccharomyces cerevisiae homologue Tof1. This study addresses the roles of Swi1 and Swi3 during a replication process perturbed by the presence of template bases alkylated by methyl methanesulfonate (MMS). Both the swi1 and swi3 mutations have additive effects on MMS sensitivity and on the MMS-induced damage checkpoint response when combined with chk1 and cds1, but they are nonadditive with hsk1. Cells with swi1, swi3, or hsk1 mutations are also defective in slowing progression through S phase in response to MMS damage. Moreover, swi1 and swi3 strains show increased levels of genomic instability even in the absence of exogenously induced DNA damage. Chromosome fragmentation, increased levels of singlestranded DNA, increased recombination, and instability of replication forks stalled in the presence of hydroxyurea are observed, consistent with the possibility that the replication process is affected in these mutants. In conclusion, Swi1, Swi3, and Hsk1 act in a novel S-phase checkpoint pathway that contributes to replication fork maintenance and to survival of alkylation damage.

Evolutionary experiments with microbes are a powerful tool to study mutations and natural selection. These experiments, however, are often limited to the well-mixed environments of a test tube or a chemostat. Since spatial organization can significantly affect evolutionary dynamics, the need is growing for evolutionary experiments in spatially structured environments. The surface of a Petri dish provides such an environment, but a more detailed understanding of microbial growth on Petri dishes is necessary to interpret such experiments. We formulate a simple deterministic reaction-diffusion model, which successfully predicts the spatial patterns created by two competing species during colony expansion. We also derive the shape of these patterns analytically without relying on microscopic details of the model. In particular, we find that the relative fitness of two microbial strains can be estimated from the logarithmic spirals created by selective sweeps. The theory is tested with strains of the budding yeast Saccharomyces cerevisiae, for spatial competitions with different initial conditions and for a range of relative fitnesses. The reaction-diffusion model also connects the microscopic parameters like growth rates and diffusion constants with macroscopic spatial patterns and predicts the relationship between fitness in liquid cultures and on Petri dishes, which we confirmed experimentally. Spatial sector patterns therefore provide an alternative fitness assay to the commonly used liquid culture fitness assays.