Higher-order Clustering Research Articles

Image Segmentation Using Higher-Order Correlation Clustering.

In this paper, a hypergraph-based image segmentation framework is formulated in a supervised manner for many high-level computer vision tasks. To consider short- and long-range dependency among various regions of an image and also to incorporate wider selection of features, a higher-order correlation clustering (HO-CC) is incorporated in the framework. Correlation clustering (CC), which is a graph-partitioning algorithm, was recently shown to be effective in a number of applications such as natural language processing, document clustering, and image segmentation. It derives its partitioning result from a pairwise graph by optimizing a global objective function such that it simultaneously maximizes both intra-cluster similarity and inter-cluster dissimilarity. In the HO-CC, the pairwise graph which is used in the CC is generalized to a hypergraph which can alleviate local boundary ambiguities that can occur in the CC. Fast inference is possible by linear programming relaxation, and effective parameter learning by structured support vector machine is also possible by incorporating a decomposable structured loss function. Experimental results on various data sets show that the proposed HO-CC outperforms other state-of-the-art image segmentation algorithms. The HO-CC framework is therefore an efficient and flexible image segmentation framework.

Inhomogeneous condensates in dilute nuclear matter and BCS-BEC crossovers

We report on recent progress in understanding pairing phenomena in low-density nuclear matter at small and moderate isospin asymmetry. A rich phase diagram has been found comprising various superfluid phases that include a homogeneous and phase-separated BEC phase of deuterons at low density and a homogeneous BCS phase, an inhomogeneous LOFF phase, and a phase-separated BCS phase at higher densities. The transition from the BEC phases to the BCS phases is characterized in terms of the evolution, from strong to weak coupling, of the condensate wavefunction and the second moment of its density distribution in r-space. We briefly discuss approaches to higher-order clustering in low-density nuclear matter.

JetpTresummation in Higgs production atNNLL′+NNLO

We present predictions for Higgs production via gluon fusion with a p_T veto on jets and with the resummation of jet-veto logarithms at NNLL'+$NNLO order. These results incorporate explicit O(alphas^2) calculations of soft and beam functions, which include the dominant dependence on the jet radius R. In particular the NNLL' order accounts for the correct boundary conditions for the N3LL resummation, for which the only unknown ingredients are higher-order anomalous dimensions. We use scale variations in a factorization theorem in both rapidity and virtuality space to estimate the perturbative uncertainties, accounting for both higher fixed-order corrections as well as higher-order towers of jet-p_T logarithms. This formalism also predicts the correlations in the theory uncertainty between the exclusive 0-jet and inclusive 1-jet bins. At the values of R used experimentally, there are important corrections due to jet algorithm clustering that include logarithms of R. Although we do not sum logarithms of R, we do include an explicit contribution in our uncertainty estimate to account for higher-order jet clustering logarithms. Precision predictions for this H+0-jet cross section and its theoretical uncertainty are an integral part of Higgs analyses that employ jet binning.

Higher virial coefficients of a unitary Fermi gas

Third and higher order quantum virial coefficients require the solution of the corresponding quantum many-body problem. Nevertheless, in an earlier paper (Phys. Rev. Lett. 108, 260402 (2012)) we proposed that the higher-order cluster integrals of a dilute unitary fermionic gas may be approximated in terms of the two-body cluster, together with an appropriate suppression factor. Although not exact, this ansatz gave a fair agreement up to fugacity z=6 with the experimentally obtained equation of state. The objective of the present note is to give some physical arguments in favor of this ansatz.

Generalized relativistic effective core potential calculations of the adiabatic potential curve and spectroscopic constants for the ground electronic state of the Ca2molecule

The potential curve, dissociation energy, equilibrium internuclear distance, and spectroscopic constants for the ground state of the Ca2 molecule are calculated with the help of the generalized relativistic effective core potential method, which allows one to exclude the inner core electrons from the calculations and to take the relativistic effects into account effectively. Extensive generalized correlation basis sets were constructed and used. The scalar relativistic coupled cluster method with corrections for high-order cluster amplitudes is used for the correlation treatment. The results are analyzed and compared with the experimental data and corresponding all-electron results. © 2013 Wiley Periodicals, Inc.

Exploring higher-order EGFR oligomerisation and phosphorylation--a combined experimental and theoretical approach.

The epidermal growth factor receptor (EGFR) kinase is generally considered to be activated by either ligand-induced dimerisation or a ligand-induced conformational change within pre-formed dimers. Ligand-induced higher-order EGFR oligomerisation or clustering has been reported but it is not clear how EGFR oligomers, as distinct from EGFR dimers, influence signaling outputs. To address this question, we combined measures of receptor clustering (microscopy; image correlation spectroscopy) and phosphorylation (Western blots) with modelling of mass-action chemical kinetics. A stable BaF/3 cell-line that contains a high proportion (>90%) of inactive dimers of EGFR-eGFP but no secreted ligand and no other detectable ErbB receptors was used as the model cell system. EGF at concentrations of greater than 1 nM was found to cluster EGFR-eGFP dimers into higher-order complexes and cause parallel increases in EGFR phosphorylation. The kinetics of EGFR clustering and phosphorylation were both rapid, plateauing within 2 minutes after stimulation with 30 nM EGF. A rule-based model was formulated to interpret the data. This model took into account ligand binding, ligand-induced conformational changes in the cytosolic tail, monomer-dimer-trimer-tetramer transitions via ectodomain- and kinase-mediated interactions, and phosphorylation. The model predicts that cyclic EGFR tetramers are the predominant phosphorylated species, in which activated receptor dimers adopt a cyclic side-by-side orientation, and that receptor kinase activation is stabilised by the intramolecular interactions responsible for cyclic tetramerization.

CD40L-Tri, a novel formulation of recombinant human CD40L that effectively activates B cells

CD40L has a well-established role in enhancing the immunostimulatory capacity of normal and malignant B cells, but a formulation suitable for clinical use has not been widely available. Like other TNF family members, in vivo and in vitro activity of CD40L requires a homotrimeric configuration, and growing evidence suggests that bioactivity depends on higher-order clustering of CD40. We generated a novel formulation of human recombinant CD40L (CD40L-Tri) in which the CD40L extracellular domain and a trimerization motif are connected by a long flexible peptide linker. We demonstrate that CD40L-Tri significantly expands normal CD19+ B cells by over 20- to 30-fold over 14 days and induces B cells to become highly immunostimulatory antigen-presenting cells (APCs). Consistent with these results, CD40L-Tri-activated B cells could effectively stimulate antigen-specific T responses (against the influenza M1 peptide) from normal volunteers. In addition, CD40L-Tri could induce malignant B cells to become effective APCs, such that tumor-directed immune responses could be probed. Together, our studies demonstrate the potent immune-stimulatory effects of CD40L-Tri on B cells that enable their expansion of antigen-specific human T cells. The potent bioactivity of CD40L-Tri is related to its ability to self-multimerize, which may be facilitated by its long peptide linker.Electronic supplementary materialThe online version of this article (doi:10.1007/s00262-012-1331-4) contains supplementary material, which is available to authorized users.

Understanding the importance of the aromatic amino-acid residues as hot-spots

Protein–protein interactions (PPI) are crucial for the establishment of life. However, its basic principles are still elusive and the recognition process is yet to be understood. It is important to look at the biomolecular structural space as a whole, in order to understand the principles behind conformation–function relationships. Since the application of an alanine scanning mutagenesis (ASM) study to the growth hormone it was demonstrated that only a small subset of residues at a protein–protein interface is essential for binding — the hot-spots (HS). Aromatic residues are some of the most typical HS at a protein–protein interface. To investigate the structural role of the interfacial aromatic residues in protein–protein interactions, we performed Molecular Dynamic (MD) simulations of protein–protein complexes in a water environment and calculated a variety of physical–chemical characteristics. ASM studies of single residues and of dimers or high-order clusters were performed to check for cooperativity within aromatic residues. Major differences were found between the behavior of non-HS aromatic residues and HS aromatic residues that can be used to design drugs to block the critical interactions or to predict major interactions at protein–protein complexes.

Spectral clustering based on local linear approximations

In the context of clustering, we assume a generative model where each cluster is the result of sampling points in the neighborhood of an embedded smooth surface; the sample may be contaminated with outliers, which are modeled as points sampled in space away from the clusters. We consider a prototype for a higher-order spectral clustering method based on the residual from a local linear approximation. We obtain theoretical guarantees for this algorithm and show that, in terms of both separation and robustness to outliers, it outperforms the standard spectral clustering algorithm (based on pairwise distances) of Ng, Jordan and Weiss (NIPS '01). The optimal choice for some of the tuning parameters depends on the dimension and thickness of the clusters. We provide estimators that come close enough for our theoretical purposes. We also discuss the cases of clusters of mixed dimensions and of clusters that are generated from smoother surfaces. In our experiments, this algorithm is shown to outperform pairwise spectral clustering on both simulated and real data.

COUNTS-IN-CYLINDERS IN THE SLOAN DIGITAL SKY SURVEY WITH COMPARISONS TON-BODY SIMULATIONS

Environmental statistics provide a necessary means of comparing the properties of galaxies in different environments and a vital test of models of galaxy formation within the prevailing, hierarchical cosmological model. We explore counts-in-cylinders, a common statistic defined as the number of companions of a particular galaxy found within a given projected radius and redshift interval. Galaxy distributions with the same two-point correlation functions do not necessarily have the same companion count distributions. We use this statistic to examine the environments of galaxies in the Sloan Digital Sky Survey, Data Release 4. We also make preliminary comparisons to four models for the spatial distributions of galaxies, based on N-body simulations, and data from SDSS DR4 to study the utility of the counts-in-cylinders statistic. There is a very large scatter between the number of companions a galaxy has and the mass of its parent dark matter halo and the halo occupation, limiting the utility of this statistic for certain kinds of environmental studies. We also show that prevalent, empirical models of galaxy clustering that match observed two- and three-point clustering statistics well fail to reproduce some aspects of the observed distribution of counts-in-cylinders on 1, 3 and 6-Mpc/h scales. All models that we explore underpredict the fraction of galaxies with few or no companions in 3 and 6-Mpc/h cylinders. Roughly 7% of galaxies in the real universe are significantly more isolated within a 6 Mpc/h cylinder than the galaxies in any of the models we use. Simple, phenomenological models that map galaxies to dark matter halos fail to reproduce high-order clustering statistics in low-density environments.

The effect of the iterative triple and quadruple cluster amplitudes on the adiabatic potential curve in the coupled cluster calculations of the ground electronic state of the Yb dimer

The dissociation energy, equilibrium internuclear distance, and spectroscopic constants for the 1Σ ground state of the Yb2 molecule are calculated. The relativistic effects are introduced through generalized relativistic effective core potentials with very high precision. The scalar relativistic coupled cluster method particularly well suited for closed-shell van der Waals systems is used for the correlation treatment. Extensive generalized correlation basis sets were constructed and used. The relatively small corrections for high-order cluster amplitudes and spin—orbit interactions are taken into account using smaller basis sets and the spin—orbit density functional theory. © 2010 Wiley Periodicals, Inc. Int J Quantum Chem, 2011

Nucleocapsid Promotes Localization of HIV-1 Gag to Uropods That Participate in Virological Synapses between T Cells

T cells adopt a polarized morphology in lymphoid organs, where cell-to-cell transmission of HIV-1 is likely frequent. However, despite the importance of understanding virus spread in vivo, little is known about the HIV-1 life cycle, particularly its late phase, in polarized T cells. Polarized T cells form two ends, the leading edge at the front and a protrusion called a uropod at the rear. Using multiple uropod markers, we observed that HIV-1 Gag localizes to the uropod in polarized T cells. Infected T cells formed contacts with uninfected target T cells preferentially via HIV-1 Gag-containing uropods compared to leading edges that lack plasma-membrane-associated Gag. Cell contacts enriched in Gag and CD4, which define the virological synapse (VS), are also enriched in uropod markers. These results indicate that Gag-laden uropods participate in the formation and/or structure of the VS, which likely plays a key role in cell-to-cell transmission of HIV-1. Consistent with this notion, a myosin light chain kinase inhibitor, which disrupts uropods, reduced virus particle transfer from infected T cells to target T cells. Mechanistically, we observed that Gag copatches with antibody-crosslinked uropod markers even in non-polarized cells, suggesting an association of Gag with uropod-specific microdomains that carry Gag to uropods. Finally, we determined that localization of Gag to the uropod depends on higher-order clustering driven by its NC domain. Taken together, these results support a model in which NC-dependent Gag accumulation to uropods establishes a preformed platform that later constitutes T-cell-T-cell contacts at which HIV-1 virus transfer occurs.

Solution NMR Investigation of the CD95/FADD Homotypic Death Domain Complex Suggests Lack of Engagement of the CD95 C Terminus

We have addressed complex formation between the death domain (DD) of the death receptor CD95 (Fas/APO-1) with the DD of immediate adaptor protein FADD using nuclear magnetic resonance (NMR) spectroscopy, mass spectrometry, and size-exclusion chromatography with in-line light scattering. We find complexation to be independent of the C-terminal 12 residues of CD95 and insensitive to mutation of residues that engage in the high-order clustering of CD95-DD molecules in a recently reported crystal structure obtained at pH 4. Differential NMR linewidths indicate that the C-terminal region of the CD95 chains remains in a disordered state and (13)C-methyl TROSY data are consistent with a lack of high degree of symmetry for the complex. The overall molecular mass of the complex is inconsistent with that in the crystal structure, and the complex dissociates at pH 4. We discuss these findings using sequence analysis of CD95 orthologs and the effect of FADD mutations on the interaction with CD95.

Link Prediction Based on Graph Topology: The Predictive Value of Generalized Clustering Coefficient

linkages among data objects is a fundamental data mining task in various application domains, including recommender systems, information retrieval, automatic Web hyperlink generation, record linkage, and communication surveillance. In many contexts link prediction is entirely based on the linkage information itself (a prominent example is the collaborative filtering recommendation). Link-structure based link prediction is closely related to a parallel and almost separate stream of research on topological modeling of large-scale graphs. Graph topological modeling builds on random graph theory to find parsimonious graph generation models reproducing empirical topological measures that summarize the global structure of a graph, such as clustering coefficient, average path length, and degree distribution. These well-studied topological measures and graph generation models have direct implications on link prediction. This paper represents initial efforts to explore the connection between link prediction and graph topology. The focus is exclusively on the predictive value of the clustering coefficient measure. The standard clustering coefficient measure is generalized to capture higher-order clustering tendencies. The proposed framework consists of a cycle formation link probability model, a procedure for estimating model parameters based on the generalized clustering coefficients, and model-based link prediction generation. Using the Enron email dataset we demonstrate that the proposed cycle formation model corresponded closely with the actual link probabilities and the link prediction algorithm based on this model outperformed existing algorithms.

Focus on the Spatial Organization of Signalling

Since the early days of light microscopy it has been apparent that cells are highly organized and compartmentalized. However, only in recent decades has the molecular basis of this spatial organization and its functional role in cellular signalling events emerged. Furthermore, modern molecular and imaging techniques, together with the intricate combination of cellular and developmental biological approaches, are extending our understanding of how individual cells respond to their environment and are organized within a tissue. Years of molecular and genetic analysis, coupled with recent advances in large‐scale approaches, may have identified many of the core components of signalling pathways, although new layers of regulation, such as small RNAs and alternative splicing, continue to add complexity. Although the parts list may soon be completed, our understanding at the level of signal integration and cross‐talk remains limited. The key to understanding cellular regulation is to complement biochemical assays in vitro with analysis on how spatial organization contributes to the properties of intra‐ and intercellular communication. This requires a real‐time quantitative analysis of endogenous components in the cellular context, which has only recently became possible with advanced microscopy techniques and biosensors. These developments open a new era in the analysis of spatial organization on different biological scales from molecules to cells to tissues and organs. The EMBO Journal covers a broad spectrum of subject areas and approaches, ranging from biochemical to cell biological, as well as those at the organism level. However, the journal has a longstanding tradition of hosting important papers on signalling, which …

High-resolution imaging of ion conductivity of Nafion ® membranes with electrochemical atomic force microscopy

Conductive electrochemical AFM images demonstrating the complex nature and structure of Nafion surface conductivity are presented. Nanoscale regions with high currents determining the overall total membrane current can be distinguished from majority domains with lower currents and non-conductive areas. The different conductive domains form ordered structures and show a specific dynamic behaviour. These observations were compared to the structural and electrical models in the literature. None of the models is able to explain all aspects of the current images. The existence of inverted micelles seems to be quite probable since the formation of agglomerates like chains and larger ordered clusters is clearly visible. This aspect is best described by the model of Schmidt-Rohr and Chen. In addition, the highly dynamic behaviour and distribution of conductive channels of Nafion leading to the formation of new current pathways also indicates the formation of different meso-phases with a high local fluctuation rate. The other discussed models also predict structural features which are in agreement with our observations like the formation of super-structures and agglomeration of fibers. The structural characterisation reflects the situation at or near the membrane surface and might differ from the bulk structure since the surface energy may have a large influence on the formation of structures during the membrane solidification process. The quite large dynamics of conductivity changes of Nafion reflected in the formation of new current pathways even at room temperature leads to the assumption that the internal structure of Nafion is subject to significant changes due to humidity and temperature variations. The local variation of individual structures may reflect the variation of concentration of hydrophilic and hydrophobic groups during membrane solidification. The minimization of surface free energy during a self-assembling process is essential for the formation of different phases and subsequent structures like chains, etc. as well as higher order clustering.

“Inquiring Love of This World”: An Implicit Love Theory of Chinese University Students

This paper reports a preliminary descriptive approach to the representation of the concept of love. Based on Spreading Activation Model, the word of love was presented as a stimulus to which 278 college students were asked to respond with at least 15 words/phrases that came to mind. Then top 100 love-related words/phrases with frequency above 4.3% were collected as units of analysis. Based on the interrelations among those words/phrases, a complete-linkage cluster analysis reached 5 high-order clusters, i.e. the five facets of the representation of love concept: 1) ethics and responsibilities; 2) romantic experience; 3) conflict and pain; 4) rationality; 5) friendly care. The results show that word associations can fundamentally reflect some differences in cognition and culture of love. Compared with Sternberg’s triangular theory of love, the findings show differences in two clusters: “conflict and pain” and “rationality”, which are not included in the triangle of love. More importantly, this research contributes to literatures on Chinese people’s love and will also inspire future research with some meaningful results.

A genome-scale analysis of the cis-regulatory circuitry underlying sonic hedgehog-mediated patterning of the mammalian limb

Sonic hedgehog (Shh) signals via Gli transcription factors to direct digit number and identity in the vertebrate limb. We characterized the Gli-dependent cis-regulatory network through a combination of whole-genome chromatin immunoprecipitation (ChIP)-on-chip and transcriptional profiling of the developing mouse limb. These analyses identified approximately 5000 high-quality Gli3-binding sites, including all known Gli-dependent enhancers. Discrete binding regions exhibit a higher-order clustering, highlighting the complexity of cis-regulatory interactions. Further, Gli3 binds inertly to previously identified neural-specific Gli enhancers, demonstrating the accessibility of their cis-regulatory elements. Intersection of DNA binding data with gene expression profiles predicted 205 putative limb target genes. A subset of putative cis-regulatory regions were analyzed in transgenic embryos, establishing Blimp1 as a direct Gli target and identifying Gli activator signaling in a direct, long-range regulation of the BMP antagonist Gremlin. In contrast, a long-range silencer cassette downstream from Hand2 likely mediates Gli3 repression in the anterior limb. These studies provide the first comprehensive characterization of the transcriptional output of a Shh-patterning process in the mammalian embryo and a framework for elaborating regulatory networks in the developing limb.

Quantum many-body theory of qubit decoherence in a finite-size spin bath

Decoherence of a center spin or qubit in a spin bath is essentially determined by the many-body bath evolution. We develop a cluster-correlation expansion (CCE) theory for the spin-bath dynamics relevant to the qubit decoherence problem. A cluster-correlation term is recursively defined as the evolution of a group of bath spins divided by the cluster correlations of all the subgroups. This correlation accounts for the authentic (nonfactorizable) collective excitations within a given group. The bath propagator is the product of all possible cluster correlation terms. For a finite-time evolution as in the qubit decoherence problem, a convergent result can be obtained by truncating the expansion up to a certain cluster size. The two-spin cluster truncation of the CCE corresponds to the pair-correlation approximation developed previously [W. Yao et al., Phys. Rev. B 74, 195301 (2006)]. In terms of the standard linked cluster expansion, a cluster-correlation term is the infinite summation of all the connected diagrams with all and only the spins in the group flip-flopped, and thus the expansion is exact whenever convergence occurs. When the individual contribution of each higher-order correlation term to the decoherence is small (while all the terms combined in product could still contribute substantially), as the usual case for relatively large baths, where the decoherence could complete well within the bath spin flip-flop time, the CCE coincides with the cluster expansion [W. M. Witzel and S. Das Sarma, Phys. Rev. B 74, 035322 (2006)]. For small baths, however, the qubit decoherence may not complete within the bath spin flip-flop time scale and thus individual higher-order cluster correlations could become significant. In such cases, only the CCE converges to the exact coherent dynamics of multispin clusters. We check the accuracy of the CCE in an exactly solvable spin-chain model.

Dynamics of interacting Brownian particles: A diagrammatic formulation

We present a diagrammatic formulation of a theory for the time dependence of density fluctuations in equilibrium systems of interacting Brownian particles. To facilitate derivation of the diagrammatic expansion, we introduce a basis that consists of orthogonalized many-particle density fluctuations. We obtain an exact hierarchy of equations of motion for time-dependent correlations of orthogonalized density fluctuations. To simplify this hierarchy we neglect contributions to the vertices from higher-order cluster expansion terms. An iterative solution of the resulting equations can be represented by diagrams with three- and four-leg vertices. We analyze the structure of the diagrammatic series for the time-dependent density correlation function and obtain a diagrammatic interpretation of reducible and irreducible memory functions. The one-loop self-consistent approximation for the latter function coincides with mode-coupling approximation for Brownian systems that was derived previously using a projection operator approach.