Indirect reference interval estimation using a convolutional neural network with application to cancer antigen 125.

Indirect methods for reference interval (RI) estimation, which use data acquired from routine pathology testing, have the potential to accelerate the establishment of RIs to account for variables such as gender and age to improve clinical assessments. However, they require more sophisticated methods of analysis due to the potential influence of pathological patients in raw clinical datasets. In this paper we develop a novel convolutional neural network (CNN) model trained on synthetic data to identify underlying healthy distributions within pathological admixtures. We present both the methodology to generate synthetic data and the CNN model. We evaluate the CNN using two synthetic test datasets, including samples from a proposed benchmark for indirect methods (RIBench) and show significant improvements compared to the reported state-of-the-art method based on the benchmark (refineR). We also demonstrate a real-world application of the model, estimating age-specific RIs for cancer antigen 125 (CA-125), a crucial biomarker for ovarian cancer diagnostics. Our results show that CA-125 RIs are strongly age-dependent, which could have important diagnostic consequences.

Unsupervised machine learning method for indirect estimation of reference intervals for chronic kidney disease in the Puerto Rican population.

Reference intervals (RIs) for clinical laboratory values are extremely important for diagnostics and treatment of patients. However, the determination of these ranges is costly and time-consuming. As a result, often different unverified RIs are used in practice for the same analyte and the same range is used for all patients despite evidence that the values are gender, age, and ethnicity dependent. Moreover, the abnormal flags are rudimentary, merely indicating if a value is within the RI. At the same time, clinical lab data generated in the everyday medical practice contains a wealth of information, that given the correct methodology, can help determine the RIs for each specific segment of the population, including populations that suffer from health disparities. In this work, we develop unsupervised machine learning methods, based on Gaussian mixtures, to determine RIs of analytes related to chronic kidney disease, using millions of routine lab results for the Puerto Rican population. We show that the measures are both gender and age dependent and we find evidence for normal age-related organ function deterioration and failure. We also show that the joint distribution of measures improves the diagnostic value of the lab results.

Organic multiferroic tunnel junctions with ferroelectric poly(vinylidene fluoride) barriers.

Organic materials are promising for applications in spintronics due to their long spin-relaxation times in addition to their chemical flexibility and relatively low production costs. Most studies of organic materials for spintronics focus on nonpolar dielectrics or semiconductors, serving as passive elements in spin transport devices. Here, we demonstrate that employing organic ferroelectrics, such as poly(vinylidene fluoride) (PVDF), as barriers in magnetic tunnel junctions (MTJs) allows new functionality in controlling the tunneling spin polarization via the ferroelectric polarization of the barrier. Using first-principles methods based on density functional theory we investigate the spin-resolved conductance of Co/PVDF/Co and Co/PVDF/Fe/Co MTJs as model systems. We show that these tunnel junctions exhibit multiple resistance states associated with different magnetization configurations of the electrodes and ferroelectric polarization orientations of the barrier. Our results indicate that organic ferroelectrics may open a new and promising route in organic spintronics with implications for low-power electronics and nonvolatile data storage.

Approaching an organic semimetal: Electron pockets at the Fermi level for a p-benzoquinonemonoimine zwitterion.

There is compelling evidence of electron pockets, at the Fermi level, in the band structure for an organic zwitterion molecule of the p-benzoquinonemonoimine type. The electronic structure of the zwitterion molecular film has a definite, although small, density of states evident at the Fermi level as well as a nonzero inner potential and thus is very different from a true insulator. In spite of a small Brillouin zone, significant band width is observed in the intermolecular band dispersion. The results demonstrate that Bloch's theorem applies to the wave vector dependence of the electronic band structure formed from the molecular orbitals of adjacent molecules in a molecular thin film of a p-benzoquinonemonoimine type zwitterion.

On the Origin of Biomolecular Networks.

Biomolecular networks have already found great utility in characterizing complex biological systems arising from pairwise interactions amongst biomolecules. Here, we explore the important and hitherto neglected role of information asymmetry in the genesis and evolution of such pairwise biomolecular interactions. Information asymmetry between sender and receiver genes is identified as a key feature distinguishing early biochemical reactions from abiotic chemistry, and a driver of network topology as biomolecular systems become more complex. In this context, we review how graph theoretical approaches can be applied not only for a better understanding of various proximate (mechanistic) relations, but also, ultimate (evolutionary) structures encoded in such networks from among all types of variations they induce. Among many possible variations, we emphasize particularly the essential role of gene duplication in terms of signaling game theory, whereby sender and receiver gene players accrue benefit from gene duplication, leading to a preferential attachment mode of network growth. The study of the resulting dynamics suggests many mathematical/computational problems, the majority of which are intractable yet yield to efficient approximation algorithms, when studied through an algebraic graph theoretic lens. We relegate for future work the role of other possible generalizations, additionally involving horizontal gene transfer, sexual recombination, endo-symbiosis, etc., which enrich the underlying graph theory even further.

Spin-transfer torque in multiferroic tunnel junctions with composite dielectric/ferroelectric barriers.

Using model calculations, we demonstrate a very high level of control of the spin-transfer torque (STT) by electric field in multiferroic tunnel junctions with composite dielectric/ferroelectric barriers. We find that, for particular device parameters, toggling the polarization direction can switch the voltage-induced part of STT between a finite value and a value close to zero, i.e. quench and release the torque. Additionally, we demonstrate that under certain conditions the zero-voltage STT, i.e. the interlayer exchange coupling, can switch sign with polarization reversal, which is equivalent to reversing the magnetic ground state of the tunnel junction. This bias- and polarization-tunability of the STT could be exploited to engineer novel functionalities such as softening/hardening of the bit or increasing the signal-to-noise ratio in magnetic sensors, which can have important implications for magnetic random access memories or for combined memory and logic devices.

DNA/RNA transverse current sequencing: intrinsic structural noise from neighboring bases.

Nanopore DNA sequencing via transverse current has emerged as a promising candidate for third-generation sequencing technology. It produces long read lengths which could alleviate problems with assembly errors inherent in current technologies. However, the high error rates of nanopore sequencing have to be addressed. A very important source of the error is the intrinsic noise in the current arising from carrier dispersion along the chain of the molecule, i.e., from the influence of neighboring bases. In this work we perform calculations of the transverse current within an effective multi-orbital tight-binding model derived from first-principles calculations of the DNA/RNA molecules, to study the effect of this structural noise on the error rates in DNA/RNA sequencing via transverse current in nanopores. We demonstrate that a statistical technique, utilizing not only the currents through the nucleotides but also the correlations in the currents, can in principle reduce the error rate below any desired precision.

Bias dependence of tunneling magnetoresistance in magnetic tunnel junctions with asymmetric barriers.

The transport properties of magnetic tunnel junctions (MTJs) are very sensitive to interface modifications. In this work we investigate both experimentally and theoretically the effect of asymmetric barrier modifications on the bias dependence of tunneling magnetoresistance (TMR) in single crystal Fe/MgO-based MTJs with (i) one crystalline and one rough interface, and (ii) with a monolayer of O deposited at the crystalline interface. In both cases we observe an asymmetric bias dependence of TMR and a reversal of its sign at large bias. We propose a general model to explain the bias dependence in these and similar systems reported earlier. The model predicts the existence of two distinct TMR regimes: (i) a tunneling regime when the interface is modified with layers of a different insulator, and (ii) a resonant regime when thin metallic layers are inserted at the interface. We demonstrate that in the tunneling regime, negative TMR is due to the high voltage which overcomes the exchange splitting in the electrodes, while the asymmetric bias dependence of TMR is due to the interface transmission probabilities. In the resonant regime, inversion of TMR could happen at zero voltage depending on the alignment of the resonance levels with the Fermi surfaces of the electrodes. Moreover, the model predicts a regime in which TMR has different signs at positive and negative bias, suggesting possibilities of combining memory with logic functions.

Ferroelectric control of magnetocrystalline anisotropy at cobalt/poly(vinylidene fluoride) interfaces.

Electric field control of magnetization is one of the promising avenues for achieving high-density energy-efficient magnetic data storage. Ferroelectric materials can be especially useful for that purpose as a source of very large switchable electric fields when interfaced with a ferromagnet. Organic ferroelectrics, such as poly(vinylidene fluoride) (PVDF), have an additional advantage of being weakly bonded to the ferromagnet, thus minimizing undesirable effects such as interface chemical modification and/or strain coupling. In this work we use first-principles density functional calculations of Co/PVDF heterostructures to demonstrate the effect of ferroelectric polarization of PVDF on the interface magnetocrystalline anisotropy that controls the magnetization orientation. We show that switching of the polarization direction alters the magnetocrystalline anisotropy energy of the adjacent Co layer by about 50%, driven by the modification of the screening charge induced by ferroelectric polarization. The effect is reduced with Co oxidation at the interface due to quenching the interface magnetization. Our results provide a new insight into the mechanism of the magnetoelectric coupling at organic ferroelectric/ferromagnet interfaces and suggest ways to achieve the desired functionality in practice.

Quantum nature of two-dimensional electron gas confinement at LaAlO3/SrTiO3 interfaces.

We perform density functional calculations to understand the mechanism controlling the confinement width of the two-dimensional electron gas (2DEG) at LaAlO_{3}/SrTiO_{3} interfaces. We find that the 2DEG confinement can be explained by the formation of metal induced gap states (MIGS) in the band gap of SrTiO3. These states are formed as the result of quantum-mechanical tunneling of the charge created at the interface due to electronic reconstruction. The attenuation length of the MIGS into the insulator is controlled by the lowest-decay-rate evanescent states of SrTiO3, as determined by its complex band structure. Our calculations predict that the 2DEG is confined in SrTiO3 within about 1 nm at the interface.

Magnetic tunnel junctions with ferroelectric barriers: prediction of four resistance States from first principles.

Magnetic tunnel junctions (MTJs), composed of two ferromagnetic electrodes separated by a thin insulating barrier layer, are currently used in spintronic devices, such as magnetic sensors and magnetic random access memories. Recently, driven by demonstrations of ferroelectricity at the nanoscale, thin-film ferroelectric barriers were proposed to extend the functionality of MTJs. Due to the sensitivity of conductance to the magnetization alignment of the electrodes (tunneling magnetoresistance) and the polarization orientation in the ferroelectric barrier (tunneling electroresistance), these multiferroic tunnel junctions (MFTJs) may serve as four-state resistance devices. On the basis of first-principles calculations, we demonstrate four resistance states in SrRuO(3)/BaTiO(3)/SrRuO(3) MFTJs with asymmetric interfaces. We find that the resistance of such a MFTJ is significantly changed when the electric polarization of the barrier is reversed and/or when the magnetizations of the electrodes are switched from parallel to antiparallel. These results reveal the exciting prospects of MFTJs for application as multifunctional spintronic devices.

Surface magnetoelectric effect in ferromagnetic metal films.

A surface magnetoelectric effect is revealed by density-functional calculations that are applied to ferromagnetic Fe(001), Ni(001), and Co(0001) films in the presence of an external electric field. The effect originates from spin-dependent screening of the electric field which leads to notable changes in the surface magnetization and the surface magnetocrystalline anisotropy. These results are of considerable interest in the area of electrically controlled magnetism and magnetoelectric phenomena.