Bayesian variable selection in structured high-dimensional covariate spaces with applications in genomics

We consider the problem of variable selection in regression modeling in high-dimensional spaces where there is known structure among the covariates. This is an unconventional variable selection problem for two reasons: (1) The dimension of the covariate space is comparable, and often much larger, than the number of subjects in the study, and (2) the covariate space is highly structured, and in some cases it is desirable to incorporate this structural information in to the model building process. We approach this problem through the Bayesian variable selection framework, where we assume that the covariates lie on an undirected graph and formulate an Ising prior on the model space for incorporating structural information. Certain computational and statistical problems arise that are unique to such high-dimensional, structured settings, the most interesting being the phenomenon of phase transitions. We propose theoretical and computational schemes to mitigate these problems. We illustrate our methods on two different graph structures: the linear chain and the regular graph of degree k. Finally, we use our methods to study a specific application in genomics: the modeling of transcription factor binding sites in DNA sequences. © 2010 American Statistical Association.

A genetically engineered thermally responsive sustained release curcumin depot to treat neuroinflammation.

Radiculopathy, a painful neuroinflammation that can accompany intervertebral disc herniation, is associated with locally increased levels of the pro-inflammatory cytokine tumor necrosis factor alpha (TNFα). Systemic administration of TNF antagonists for radiculopathy in the clinic has shown mixed results, and there is growing interest in the local delivery of anti-inflammatory drugs to treat this pathology as well as similar inflammatory events of peripheral nerve injury. Curcumin, a known antagonist of TNFα in multiple cell types and tissues, was chemically modified and conjugated to a thermally responsive elastin-like polypeptide (ELP) to create an injectable depot for sustained, local delivery of curcumin to treat neuroinflammation. ELPs are biopolymers capable of thermally-triggered in situ depot formation that have been successfully employed as drug carriers and biomaterials in several applications. ELP-curcumin conjugates were shown to display high drug loading, rapidly release curcumin in vitro via degradable carbamate bonds, and retain in vitro bioactivity against TNFα-induced cytotoxicity and monocyte activation with IC50 only two-fold higher than curcumin. When injected proximal to the sciatic nerve in mice via intramuscular (i.m.) injection, ELP-curcumin conjugates underwent a thermally triggered soluble-insoluble phase transition, leading to in situ formation of a depot that released curcumin over 4days post-injection and decreased plasma AUC 7-fold.

Microwave modelling and validation in food thawing applications

Developing temperature fields in frozen cheese sauce undergoing microwave heating were simulated and measured. Two scenarios were investigated: a centric and offset placement on the rotating turntable. Numerical modeling was performed using a dedicated electromagnetic Finite Difference Time Domain (FDTD) module that was two-way coupled to the PHYSICA multiphysics package. Two meshes were used: the food material and container were meshed for the heat transfer and the microwave oven cavity and waveguide were meshed for the microwave field. Power densities obtained on the structured FDTD mesh were mapped onto the unstructured finite volume method mesh for each time-step/turntable position. On heating for each specified time-step the temperature field was mapped back onto the FDTD mesh and the electromagnetic properties were updated accordingly. Changes in thermal/electric properties associated with the phase transition were fully accounted for as well as heat losses from product to cavity. Detailed comparisons were carried out for the centric and offset placements, comparing experimental temperature profiles during microwave thawing with those obtained by numerical simulation.

Numerical simulation of microwave thawing using a coupled solver approach

Thawing of a frozen food product in a domestic microwave oven is numerically simulated using a coupled solver approach. The approach consists of a dedicated electromagnetic FDTD solver and a closely coupled UFVM multi-physics package. Two overlapping numerical meshes are defined; the food material and container were meshed for heat transfer and phase change solution, whilst the microwave oven cavity and waveguide were meshed for the microwave irradiation. The two solution domains were linked using a cross-mapping routine. This approach allowed the rotation of the food load to be captured. Power densities obtained on the structured FDTD mesh were interpolated onto the UFVM mesh for each timestep/turntable position. The UFVM solver utilised the power density data to advance the temperature and phase distribution solution. The temperature-dependant dielectric and thermo-physical properties of the food load were updated prior to revising the electromagnetic solution. Changes in thermal/electric properties associated with the phase transition were fully accounted for as well as heat losses from product to cavity. Two scenarios were investigated: a centric and eccentric placement on the turntable. Developing temperature fields predicted by the numerical solution are validated against experimentally obtained data. Presented results indicate the feasibility of fully coupled simulations of the microwave heating of a frozen product. (© 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim)

Density functional and Monte Carlo studies of sulfur. I. Structure and bonding in S-n rings and chains (n=2-18) (vol 118, pg 9257, 2003)

Density functional calculations have been performed for ring isomers of sulfur with up to 18 atoms, and for chains with up to ten atoms. There are many isomers of both types, and the calculations predict the existence of new forms. Larger rings and chains are very flexible, with numerous local energy minima. Apart from a small, but consistent overestimate in the bond lengths, the results reproduce experimental structures where known. Calculations are also performed on the energy surfaces of S8 rings, on the interaction between a pair of such rings, and the reaction between one S8 ring and the triplet diradical S8 chain. The results for potential energies, vibrational frequencies, and reaction mechanisms in sulfur rings and chains provide essential ingredients for Monte Carlo simulations of the liquid–liquid phase transition. The results of these simulations will be presented in Part II.

First-principles study of ferroelectricity and isotope effects in H-bonded KH2PO4 crystals

By means of extensive first-principles calculations we studied the ferroelectric phase transition and the associated isotope effect in KH2PO4 (KDP). Our calculations revealed that the spontaneous polarization of the ferroelectric phase is due to electronic charge redistributions and ionic displacements which are a consequence of proton ordering, and not vice versa. The experimentally observed double-peaked proton distribution in the paraelectric phase cannot be explained by a dynamics of only protons. This requires, instead, collective displacements within clusters that include also the heavier ions. These tunneling clusters can explain the recent evidence of tunneling obtained from Compton scattering measurements. The sole effect of mass change upon deuteration is not sufficient to explain the huge isotope effect. Instead, we find that structural modifications deeply connected with the chemistry of the H bonds produce a feedback effect on tunneling that strongly enhances the phenomenon. The resulting influence of the geometric changes on the isotope effect agrees with experimental data from neutron scattering. Calculations under pressure allowed us to analyze the issue of universality in the disappearance of ferroelectricity upon compression. Compressing DKDP so that the distance between the two peaks in the deuteron distribution is the same as for protons in KDP, corresponds to a modification of the underlying double-well potential, which becomes 23 meV shallower. This energy difference is what is required to modify the O-O distance in such a way as to have the same distribution for protons and deuterons. At the high pressures required experimentally, the above feedback mechanism is crucial to explain the magnitude of the geometrical effect.

Nanoscale ferroelectrics machined from single crystals

This paper summarises some of the most recent work that has been done on nanoscale ferroelectrics as a result of a joint collaborative research effort involving groups in Queen's University Belfast, the University of Cambridge and the University of St. Andrews. Attempts have been made to observe fundamental effects of reduced size, and increasing morphological complexity, on ferroelectric behaviour by studying the functional response and domain characteristics in nanoscale single crystal material, whose size and morphology have been defined by Focused Ion Beam (FIB) patterning. This approach to nanoshape fabrication has allowed the following broad statements to be made: (i) in single crystal BaTiO3 sheets, permittivity and phase transition behaviour is not altered from that of bulk material down to a thickness of similar to 75 nm; (ii) in single crystal BaTiO3 sheets and nanowires changes in observed domain morphologies are consistent with large scale continuum modeling.

Understanding thickness effects in thin film capacitors

Pulsed Laser Deposition (PLD) was used to make Au/(Ba0.5Sr0.5)TiO3/(La0.5Sr0.5) CoO3/MgO thin film capacitor structures. Functional properties were studied with changing BST thickness from similar to1265 nm to similar to63 nm. The dielectric constant was found to decrease, and migration of T-m (the temperature at which the dielectric constant is maximum) to lower temperatures occurred as thickness was reduced. Curie-Weiss plots of the as-obtained dielectric data, indicated that the Curie temperature was also systemmatically progressively depressed. Further, fitting to expressions previously used to describe diffuse phase transitions suggested increased diffuseness in transformation behaviour as film thickness decreased. This paper discusses the care needed in interpreting the observations given above. We make particular distinction between the apparent Curie-temperature derived from Curie-Weiss plots of as-measured data, and the inherent Curie temperature determined after correction for the interfacial capacitance. We demonstrate that while the apparent Curie temperature decreases as thickness decreases, the inherent Curie temperature is thickness independent. Thickness-invariant phase transition behaviour is confirmed from analysis of polarisation loops, and from examination of the temperature dependence of the loss-tangent. We particularly note that correction of data for interfacial capacitance does not alter the position of T-m. We must therefore conclude that the position of T-m is not related simply to phase transformation behaviour in BST thin films.

Origin of Ferroelastic domains in Free-Standing Single Crystal Ferroelectric Films

The origin of the unusual 90 degrees ferroelectric/ferroelastic domains, consistently observed in recent studies on mesoscale and nanoscale free-standing single crystals of BaTiO3 [Schilling , Phys. Rev. B 74, 024115 (2006); Schilling , Nano Lett. 7, 3787 (2007)], has been considered. A model has been developed which postulates that the domains form as a response to elastic stress induced by a surface layer which does not undergo the paraelectric-ferroelectric cubic-tetragonal phase transition. This model was found to accurately account for the changes in domain periodicity as a function of size that had been observed experimentally. The physical origin of the surface layer might readily be associated with patterning damage, seen in experiment; however, when all evidence of physical damage is removed from the BaTiO3 surfaces by thermal annealing, the domain configuration remains practically unchanged. This suggests a more intrinsic origin, such as the increased importance of surface tension at small dimensions. The effect of surface tension is also shown to be proportional to the difference in hardness between the surface and the interior of the ferroelectric. The present model for surface-tension induced twinning should also be relevant for finely grained or core-shell structured ceramics.

The Influence of Point Defects and Inhomogeneous Strain on the Functional Behaviour of Thin film Ferroelectrics

An attempt has been made to unequivocally identify the influence that inhomogeneous strain fields, surrounding point defects, have on the functional properties of thin film ferroelectrics. Single crystal thin film lamellae of BaTiO3 have been integrated into capacitor structures, and the functional differences between those annealed in oxygen and those annealed in nitrogen have been mapped. Key features, such as the change in the paraelectric-ferroelectric phase transition from first to second order were noted and found to be consistent with mean field modeling predictions for the effects of inhomogeneous strain. Switching characteristics appeared to be unaffected, suggesting that point defects have a low efficacy in domain wall pinning.

Domains in Ferroelectric Nanodots

Almost free-standing single crystal mesoscale and nanoscale dots of ferroelectric BaTiO3 have been made by direct focused ion beam patterning of bulk single crystal material. The domain structures which appear in these single crystal dots, after cooling through the Curie temperature, were observed to form into quadrants, with each quadrant consisting of fine 90° stripe domains. The reason that these rather complex domain configurations form is uncertain, but we consider and discuss three possibilities for their genesis: first, that the quadrant features initially form to facilitate field-closure, but then develop 90° shape compensating stripe domains in order to accommodate disclination stresses; second, that they are the result of the impingement of domain packets which nucleate at the sidewalls of the dots forming “Forsbergh” patterns (essentially the result of phase transition kinetics); and third, that 90° domains form to conserve the shape of the nanodot as it is cooled through the Curie temperature but arrange into quadrant packets in order to minimize the energy associated with uncompensated surface charges (thus representing an equilibrium state). While the third model is the preferred one, we note that the second and third models are not mutually exclusive.

Temperature-Driven Mixing-Demixing Behavior of Binary Mixtures of the Ionic Liquid Choline Bis(trifluoromethylsulfonyl)imide and Water

The ionic liquid (2-hydroxyethylammonium)trimethylammonium) bis(trifluoromethylsulfonyl)imide (choline bistriflimide) was obtained as a supercooled liquid at room temperature (melting point = 30 degrees C). Crystals of choline bistriflimide suitable for structure determination were grown from the melt in situ on the X-ray diffractometer. The choline cation adopts a folded conformation, whereas the bistriflimide anion exhibits a transoid conformation. The choline cation and the bistriflimide anion are held together by hydrogen bonds between the hydroxyl proton and a sulfonyl oxygen atom. This hydrogen bonding is of importance for the temperature-dependent solubility proper-ties of the ionic liquid. Choline bistriflimide is not miscible with water at room temperature, but forms one phase with water at temperatures above 72 degrees C (equals upper critical solution temperature). H-1 NMR studies show that the hydrogen bonds between the choline cation and the bistriflimide anion are substantially weakened above this temperature. The thermophysical properties of water-choline bistriflimide binary mixtures were furthermore studied by a photopyroelectric technique and by adiabatic scanning calorimetry (ASC). By photothermal analysis, besides highly accurate values for the thermal conductivity and effusivity of choline bistriflimide at 30 degrees C, the detailed temperature dependence of both the thermal conductivity and effusivity of the upper and lower part of a critical water-choline bistriflimide mixture in the neighborhood of the mixing-demixing phase transition could be determined with high resolution and accuracy. Together with high resolution ASC data for the heat capacity, experimental values were obtained for the critical exponents alpha and beta, and for the critical amplitude ratio G(+)/G(-). These three values were found to be consistent with theoretical expectations for a three dimensional Ising-type of critical behavior of binary liquid mixtures.

Quantitative mapping of aqueous microfluidic temperature with sub-degree resolution using fluorescence lifetime imaging microscopy

The use of a water-soluble, thermo-responsive polymer as a highly sensitive fluorescence-lifetime probe of microfluidic temperature is demonstrated. The fluorescence lifetime of poly(N-isopropylacrylamide) labelled with a benzofurazan fluorophore is shown to have a steep dependence on temperature around the polymer phase transition and the photophysical origin of this response is established. The use of this unusual fluorescent probe in conjunction with fluorescence lifetime imaging microscopy (FLIM) enables the spatial variation of temperature in a microfluidic device to be mapped, on the micron scale, with a resolution of less than 0.1 degrees C. This represents an increase in temperature resolution of an order of magnitude over that achieved previously by FLIM of temperature-sensitive dyes

A study of the relationship between process conditions and mechanical strength of mineralized red algae in the preparation of a marine-derived bone void filler

Bone void fillers that can enhance biological function to augment skeletal repair have significant therapeutic potential in bone replacement surgery. This work focuses on the development of a unique microporous (0.5-10 mu m) marine-derived calcium phosphate bioceramic granule. It was prepared fro Corallina officinalis, a mineralized red alga, using a novel manufacturing process. This involved thermal processing, followed by a low pressure-temperature chemical synthesis reaction. The study found that the ability to maintain the unique algal morphology was dependent on the thermal processing conditions. This study investigates the effect of thermal heat treatment on the physiochemical properties of the alga. Thermogravimetric analysis was used to monitor its thermal decomposition. The resultant thermograms indicated the presence of a residual organic phase at temperatures below 500 degrees C and an irreversible solid-state phase transition from mg-rich-calcite to calcium oxide at temperatures over 850 degrees C. Algae and synthetic calcite were evaluated following heat treatment in an air-circulating furance at temperatures ranging from 400 to 800 degrees C. The highest levels of mass loss occurred between 400-500 degrees C and 700-800 degrees C, which were attributed to the organic and carbonate decomposition respectively. The changes in mechanical strength were quantified using a simple mechanical test, which measured the bulk compressive strength of the algae. The mechanical test used may provide a useful evaluation of the compressive properties of similar bone void fillers that are in granular form. The study concluded that soak temperatures in the range of 600 to 700 degrees C provided the optimum physiochemical properties as a precursor to conversion to hydroxyapatite (HA). At these temperatures, a partial phase transition to calcium oxide occurred and the original skeletal morphology of the alga remained intact.

Manipulating and protecting entanglement by means of spin environments

We study the dynamical behavior of two initially entangled qubits, each locally coupled to an environment embodied by an interacting spin chain. We consider energy-exchange qubit-environment couplings resulting in rich and highly non-trivial entanglement dynamics. We obtain exact results for the time evolution of the concurrence between the two qubits and find that, by tuning the interaction parameters, one can freeze the dynamics of entanglement, therefore inhibiting their relaxation into the spin environments, as well as activate a sudden-death phenomenon. We also discuss the effects of an environmental quantum phase transition on the features of the two-qubit entanglement dynamics.