Discovering Higher-Order Interactions Through Neural Information Decomposition

If regularity in data takes the form of higher-order functions among groups of variables, models which are biased towards lower-order functions may easily mistake the data for noise. To distinguish whether this is the case, one must be able to quantify the contribution of different orders of dependence to the total information. Recent work in information theory attempts to do this through measures of multivariate mutual information (MMI) and information decomposition (ID). Despite substantial theoretical progress, practical issues related to tractability and learnability of higher-order functions are still largely unaddressed. In this work, we introduce a new approach to information decomposition—termed Neural Information Decomposition (NID)—which is both theoretically grounded, and can be efficiently estimated in practice using neural networks. We show on synthetic data that NID can learn to distinguish higher-order functions from noise, while many unsupervised probability models cannot. Additionally, we demonstrate the usefulness of this framework as a tool for exploring biological and artificial neural networks.     

Candida antarctica B Lipase-Loaded Microreactor for the Automated Derivatization of Lipids

A key bottleneck in the profiling of lipids is the multistep derivatization required prior to gas chromatography (GC) analysis. A single in-vial lipid derivatization and analysis may significantly minimize sample loss and improve analytical sensitivity. A cotton fiber-supported poly(glycidylmethacrylate-co-ethylene glycol dimethacrylate) polymer microbrush microreactor loaded with Candida antarctica lipase B was developed for the facile conversion of triacylglycerols into fatty acid ethyl ester derivatives for gas chromatograph–mass spectrometry (GC–MS) analysis. The polymer microbrush microreactor was fabricated in effort to provide efficient, simplified, cost effective, and high-throughput GC–MS determination of triacylglycerols. The polymer microbrush microreactor was used as an in-vial triacylglycerol transesterification platform, with economical sample consumption of less than or equal to 100?µL and significant reduction of reagents. To evaluate the polymer microbrush microreactor performance for lipids, a triolein standard and camelina oil triacylglycerols were quantitatively transformed into ethyl oleate and fatty acid ethyl esters, respectively, following a 3?h reaction time. The lipase-loaded cotton fiber-supported poly(glycidylmethacrylate-co-ethylene glycol dimethacrylate) polymer microbrush microreactors were reusable for up to five times for quantitative transesterification with minimal loss of lipase activity.     

Outer-sphere effects on reduction potentials of copper sites in proteins: the curious case of high potential type 2 C112D/M121E Pseudomonas aeruginosa azurin

Redox and spectroscopic (electronic absorption, multifrequency electron paramagnetic resonance (EPR), and X-ray absorption) properties together with X-ray crystal structures are reported for the type 2 Cu(II) C112D/M121E variant of Pseudomonas aeruginosa azurin. The results suggest that Cu(II) is constrained from interaction with the proximal glutamate; this structural frustration implies a "rack" mechanism for the 290 mV (vs NHE) reduction potential measured at neutral pH. At high pH (～9), hydrogen bonding in the outer coordination sphere is perturbed to allow axial glutamate ligation to Cu(II), with a decrease in potential to 119 mV. These results highlight the role played by outer-sphere interactions, and the structural constraints they impose, in determining the redox behavior of transition metal protein cofactors.     

Validation of an LC/MS method for the determination of gemfibrozil in human plasma and its application to a pharmacokinetic study

Gemfibrozil, a fibric acid hypolipidemic agent, is increasingly being used in clinical drug–drug interaction studies as an inhibitor of drug metabolizing enzymes and drug transporters. The validation of a fast, accurate and precise LC/MS method is described for the quantitative determination of gemfibrozil in an EDTA‐anticoagulated human plasma matrix. Briefly, gemfibrozil was extracted from human plasma by an acetonitrile protein precipitation method. The assay was reproducible with intra‐assay precision between 1.6 and 10.7%, and inter‐assay precision ranging from 4.4 to 7.8%. The assay also showed good accuracy, with intra‐assay concentrations within 85.6–108.7% of the expected value, and inter‐assay concentrations within 89.4–104.0% of the expected value. The linear concentration range was between 0.5 and 50 µg/mL with a lower limit of quantitation of 0.5 µg/mL when 125 µL of plasma were extracted. This LC/MS method yielded a quick, simple and reliable protocol for determining gemfibrozil concentrations in plasma and is applicable to clinical pharmacokinetic studies. Copyright © 2010 John Wiley & Sons, Ltd.     

Asymptotic formulae for likelihood-based tests of new physics

We describe likelihood-based statistical tests for use in high energy physics for the discovery of new phenomena and for construction of confidence intervals on model parameters. We focus on the properties of the test procedures that allow one to account for systematic uncertainties. Explicit formulae for the asymptotic distributions of test statistics are derived using results of Wilks and Wald. We motivate and justify the use of a representative data set, called the “Asimov data set”, which provides a simple method to obtain the median experimental sensitivity of a search or measurement as well as fluctuations about this expectation.     

Measurements of degree of sensitization (DoS) in aluminum alloys using EMAT ultrasound

Sensitization in 5XXX aluminum alloys is an insidious problem characterized by the gradual formation and growth of beta phase (Mg₂Al₃) at grain boundaries, which increases the susceptibility of alloys to intergranular corrosion (IGC) and intergranular stress-corrosion cracking (IGSCC). The degree of sensitization (DoS) is currently quantified by the ASTM G67 Nitric Acid Mass Loss Test, which is destructive and time consuming. A fast, reliable, and non-destructive method for rapid detection and the assessment of the condition of DoS in AA5XXX aluminum alloys in the field is highly desirable. In this paper, we describe a non-destructive method for measurements of DoS in aluminum alloys with an electromagnetic acoustic transducer (EMAT). AA5083 aluminum alloy samples were sensitized at 100 °C with processing times varying from 7 days to 30 days. The DoS of sensitized samples was first quantified with the ASTM 67 test in the laboratory. Both ultrasonic velocity and attenuation in sensitized specimens were then measured using EMAT and the results were correlated with the DoS data. We found that the longitudinal wave velocity was almost a constant, independent of the sensitization, which suggests that the longitudinal wave can be used to determine the sample thickness. The shear wave velocity and especially the shear wave attenuation are sensitive to DoS. Relationships between DoS and the shear velocity, as well as the shear attenuation have been established. Finally, we performed the data mining to evaluate and improve the accuracy in the measurements of DoS in aluminum alloys with EMAT.     

Access to Nitriles from Aldehydes Mediated by an Oxoammonium Salt

A scalable, high yielding, rapid route to access an array of nitriles from aldehydes mediated by an oxoammonium salt (4‐acetylamino‐2,2,6,6‐tetramethylpiperidine‐1‐oxoammonium tetrafluoroborate) and hexamethyldisilazane (HMDS) as an ammonia surrogate has been developed. The reaction likely involves two distinct chemical transformations: reversible silyl‐imine formation between HMDS and an aldehyde, followed by oxidation mediated by the oxoammonium salt and desilylation to furnish a nitrile. The spent oxidant can be easily recovered and used to regenerate the oxoammonium salt oxidant.     

Incorporating high-pressure electroosmotic pump and a nano-flow gradient generator into a miniaturized liquid chromatographic system for peptide analysis

We integrate a high-pressure electroosmotic pump (EOP), a nanoflow gradient generator, and a capillary column into a miniaturized liquid chromatographic system that can be directly coupled with a mass spectrometer for proteomic analysis. We have recently developed a low-cost high-pressure EOP capable of generating pressure of tens of thousands psi, ideal for uses in miniaturized HPLC. The pump worked smoothly when it was used for isocratic elutions. When it was used for gradient elutions, generating reproducible gradient profiles was challenging; because the pump rate fluctuated when the pump was used to pump high-content organic solvents. This presents an issue for separating proteins/peptides since high-content organic solvents are often utilized. In this work, we solve this problem by incorporating our high-pressure EOP with a nano-flow gradient generator so that the EOP needs only to pump an aqueous solution. With this combination, we develop a capillary-based nano-HPLC system capable of performing nano-flow gradient elution; the pump rate is stable, and the gradient profiles are reproducible and can be conveniently tuned. To demonstrate its utility, we couple it with either a UV absorbance detector or a mass spectrometer for peptide separations.     

Giant phospholipid/block copolymer hybrid vesicles: mixing behavior and domain formation

Lipids and block copolymers can be individually assembled into unsupported, spherical membranes (liposomes or polymersomes), each having their own particular benefits and limitations. Here we demonstrate the preparation of microscale, hybrid "lipopolymersomes" composed of the common lipid POPC (1-palmitoyl-2-oleoyl-sn-glycero-3-phosphatidylcholine) and the commercially available copolymer PBd-b-PEO (polybutadiene-b-poly(ethylene oxide)) with the goal of incorporating the advantageous qualities of the unitary systems into mixed-membrane capsules. We investigate the lipopolymersomes using confocal fluorescence microscopy and demonstrate that these hybrid membranes are well mixed on nanoscopic length scales within the permittable compositional windows for hybrid vesicle formation. We measure the intramembrane dynamics and mechanical properties of these hybrid membranes by fluorescence recovery after photobleaching (FRAP) and micropipet aspiration, respectively. For the first time, we demonstrate the demixing of lipid-rich and polymer-rich membrane domains within the same vesicle membrane. This is achieved by the biotinylation of one of the constituent species and cross linking with the protein NeutrAvidin. The resultant domain patterning is dependent upon which component carries the biotin functionality: cross linking of the copolymer species results in domains that ripen into a single, large, copolymer-rich island, and cross linking of the lipids yields many small, "spot-like", lipid-rich domains within a copolymer-rich matrix. We discuss these morphological differences in terms of the fluidity and mechanical properties of the membrane phases and the possible resultant interdomain interactions within the membrane. These heterogeneous hybrid lipopolymersomes could find applications in fields such as targeted delivery, controlled release, and environmental detection assays where these capsules possess the characteristics of biocompatible lipid membranes combined with enhanced mechanical strength and stability from the copolymer matrix.