Computational learning of the conditional phase-type (C-Ph) distribution: Learning C-Ph distributions

This paper presents a new algorithm for learning the structure of a special type of Bayesian network. The conditional phase-type (C-Ph) distribution is a Bayesian network that models the probabilistic causal relationships between a skewed continuous variable, modelled by the Coxian phase-type distribution, a special type of Markov model, and a set of interacting discrete variables. The algorithm takes a dataset as input and produces the structure, parameters and graphical representations of the fit of the C-Ph distribution as output.The algorithm, which uses a greedy-search technique and has been implemented in MATLAB, is evaluated using a simulated data set consisting of 20,000 cases. The results show that the original C-Ph distribution is recaptured and the fit of the network to the data is discussed.

Design optimization of cold-formed steel portal frames taking into account the effect of building topology

Cold-formed steel portal frames are a popular form of construction for low-rise commercial, light industrial and agricultural buildings with spans of up to 20 m. In this article, a real-coded genetic algorithm is described that is used to minimize the cost of the main frame of such buildings. The key decision variables considered in this proposed algorithm consist of both the spacing and pitch of the frame as continuous variables, as well as the discrete section sizes.A routine taking the structural analysis and frame design for cold-formed steel sections is embedded into a genetic algorithm. The results show that the real-coded genetic algorithm handles effectively the mixture of design variables, with high robustness and consistency in achieving the optimum solution. All wind load combinations according to Australian code are considered in this research. Results for frames with knee braces are also included, for which the optimization achieved even larger savings in cost.

Membrane Topology and Identification of Critical Amino Acid Residues in the Wzx O-Antigen Translocase from Escherichia coli O157:H4

Wzx belongs to a family of membrane proteins involved in the translocation of isoprenoid lipid-linked glycans, which is loosely related to members of the major facilitator superfamily. Despite Wzx homologs performing a conserved function, it has been difficult to pinpoint specific motifs of functional significance in their amino acid sequences. Here, we elucidate the topology of the Escherichia coli O157 Wzx (Wzx(EcO157)) by a combination of bioinformatics and substituted cysteine scanning mutagenesis, as well as targeted deletion-fusions to green fluorescent protein and alkaline phosphatase. We conclude that Wzx(EcO157) consists of 12 transmembrane (TM) helices and six periplasmic and five cytosolic loops, with N and C termini facing the cytoplasm. Four TM helices (II, IV, X, and XI) contain polar residues (aspartic acid or lysine), and they may form part of a relatively hydrophilic core. Thirty-five amino acid replacements to alanine or serine were targeted to five native cysteines and most of the aspartic acid, arginine, and lysine residues. From these, only replacements of aspartic acid-85, aspartic acid-326, arginine-298, and lysine-419 resulted in a protein unable to support O-antigen production. Aspartic acid-85 and lysine-419 are located in TM helices II and XI, while arginine-298 and aspartic acid-326 are located in periplasmic and cytosolic loops 4, respectively. Further analysis revealed that the charge at these positions is required for Wzx function since conservative substitutions maintaining the same charge polarity resulted in a functional protein, whereas those reversing or eliminating polarity abolished function. We propose that the functional requirement of charged residues at both sides of the membrane and in two TM helices could be important to allow the passage of the Und-PP-linked saccharide substrate across the membrane.

Validation of membrane protein topology models by oxidative labeling and mass spectrometry

Computer-assisted topology predictions are widely used to build low-resolution structural models of integral membrane proteins (IMPs). Experimental validation of these models by traditional methods is labor intensive and requires modifications that might alter the IMP native conformation. This work employs oxidative labeling coupled with mass spectrometry (MS) as a validation tool for computer-generated topology models. ·OH exposure introduces oxidative modifications in solvent-accessible regions, whereas buried segments (e.g., transmembrane helices) are non-oxidizable. The Escherichia coli protein WaaL (O-antigen ligase) is predicted to have 12 transmembrane helices and a large extramembrane domain (Pérez et al., Mol. Microbiol. 2008, 70, 1424). Tryptic digestion and LC-MS/MS were used to map the oxidative labeling behavior of WaaL. Met and Cys exhibit high intrinsic reactivities with ·OH, making them sensitive probes for solvent accessibility assays. Overall, the oxidation pattern of these residues is consistent with the originally proposed WaaL topology. One residue (M151), however, undergoes partial oxidation despite being predicted to reside within a transmembrane helix. Using an improved computer algorithm, a slightly modified topology model was generated that places M151 closer to the membrane interface. On the basis of the labeling data, it is concluded that the refined model more accurately reflects the actual topology of WaaL. We propose that the combination of oxidative labeling and MS represents a useful strategy for assessing the accuracy of IMP topology predictions, supplementing data obtained in traditional biochemical assays. In the future, it might be possible to incorporate oxidative labeling data directly as constraints in topology prediction algorithms.

Functional characterization and membrane topology of Escherichia coli WecA, a sugar-phosphate transferase initiating the biosynthesis of enterobacterial common antigen and O-antigen lipopolysaccharide

WecA is an integral membrane protein that initiates the biosynthesis of enterobacterial common antigen and O-antigen lipopolysaccharide (LPS) by catalyzing the transfer of N-acetylglucosamine (GlcNAc)-1-phosphate onto undecaprenyl phosphate (Und-P) to form Und-P-P-GlcNAc. WecA belongs to a large family of eukaryotic and prokaryotic prenyl sugar transferases. Conserved aspartic acids in putative cytoplasmic loops 2 (Asp90 and Asp91) and 3 (Asp156 and Asp159) were targeted for replacement mutagenesis with either glutamic acid or asparagine. We examined the ability of each mutant protein to complement O-antigen LPS synthesis in a wecA-deficient strain and also determined the steady-state kinetic parameters of the mutant proteins in an in vitro transfer assay. Apparent K(m) and V(max) values for UDP-GlcNAc, Mg(2+), and Mn(2+) suggest that Asp156 is required for catalysis, while Asp91 appears to interact preferentially with Mg(2+), possibly playing a role in orienting the substrates. Topological analysis using the substituted cysteine accessibility method demonstrated the cytosolic location of Asp90, Asp91, and Asp156 and provided a more refined overall topological map of WecA. Also, we show that cells expressing a WecA derivative C terminally fused with the green fluorescent protein exhibited a punctate distribution of fluorescence on the bacterial surface, suggesting that WecA localizes to discrete regions in the bacterial plasma membrane.

Computational identification of self-inhibitory peptides from envelope proteins

Fusion process is known to be the initial step of viral infection and hence targeting the entry process is a promising strategy to design antiviral therapy. The self-inhibitory peptides derived from the enveloped (E) proteins function to inhibit the proteinprotein interactions in the membrane fusion step mediated by the viral E protein. Thus, they have the potential to be developed into effective antiviral therapy. Herein, we have developed a Monte Carlo-based computational method with the aim to identify and optimize potential peptide hits from the E proteins. The stability of the peptides, which indicates their potential to bind in situ to the E proteins, was evaluated by two different scoring functions, dipolar distance-scaled, finite, ideal-gas reference state and residue-specific all-atom probability discriminatory function. The method was applied to a-helical Class I HIV-1 gp41, beta-sheet Class II Dengue virus (DENV) type 2 E proteins, as well as Class III Herpes Simplex virus-1 (HSV-1) glycoprotein, a E protein with a mixture of a-helix and beta-sheet structural fold. The peptide hits identified are in line with the druggable regions where the self-inhibitory peptide inhibitors for the three classes of viral fusion proteins were derived. Several novel peptides were identified from either the hydrophobic regions or the functionally important regions on Class II DENV-2 E protein and Class III HSV-1 gB. They have potential to disrupt the proteinprotein interaction in the fusion process and may serve as starting points for the development of novel inhibitors for viral E proteins. 

The Computational Complexity of Dominance and Consistency in CP-Nets

We investigate the computational complexity of testing dominance and consistency in CP-nets. Previously, the complexity of dominance has been determined for restricted classes in which the dependency graph of the CP-net is acyclic. However, there are preferences of interest that define cyclic dependency graphs; these are modeled with general CP-nets. In our main results, we show here that both dominance and consistency for general CP-nets are PSPACE-complete. We then consider the concept of strong dominance, dominance equivalence and dominance incomparability, and several notions of optimality, and identify the complexity of the corresponding decision problems. The reductions used in the proofs are from STRIPS planning, and thus reinforce the earlier established connections between both areas.

Computational techniques for a simple theory of conditional preferences

A simple logic of conditional preferences is defined, with a language that allows the compact representation of certain kinds of conditional preference statements, a semantics and a proof theory. CP-nets and TCP-nets can be mapped into this logic, and the semantics and proof theory generalise those of CP-nets and TCP-nets. The system can also express preferences of a lexicographic kind. The paper derives various sufficient conditions for a set of conditional preferences to be consistent, along with algorithmic techniques for checking such conditions and hence confirming consistency. These techniques can also be used for totally ordering outcomes in a way that is consistent with the set of preferences, and they are further developed to give an approach to the problem of constrained optimisation for conditional preferences.

Mixing operators on spaces with weak topology

We prove that a continuous linear operator T on a topological vector space X with weak topology is mixing if and only if the dual operator T' has no finite dimensional invariant subspaces. This result implies the characterization of hypercyclic operators on the space $\omega$ due to Herzog and Lemmert and implies the result of Bayart and Matheron, who proved that for any hypercyclic operator T on $\omega$, $T\oplus T$ is also hypercyclic.

A computational study of the mechanism of CO oxidation by a ceria supported surface rhodium oxide layer

A mechanism of CO oxidation by a thin surface oxide of Rh supported on ceria is proposed: CO is oxidized by the Rh-oxide film, which is subsequently reoxidized by a ceria surface O atom. The proposed mechanism is supported by in situ Raman spectroscopic investigations.

Deployment on GPUs of an application in computational atomic physics

This paper describes the deployment on GPUs of PROP, a program of the 2DRMP suite which models electron collisions with H-like atoms and ions. Because performance on GPUs is better in single precision than in double precision, the numerical stability of the PROP program in single precision has been studied. The numerical quality of PROP results computed in single precision and their impact on the next program of the 2DRMP suite has been analyzed. Successive versions of the PROP program on GPUs have been developed in order to improve its performance. Particular attention has been paid to the optimization of data transfers and of linear algebra operations. Performance obtained on several architectures (including NVIDIA Fermi) are presented.