Average Cross Research Articles

Abstract 198: Quantifying miRNA activity in single cell clusters

Abstract Background: MicroRNAs are small noncoding RNAs that mediate gene regulation at the post-transcriptional level via multiple mechanisms such as mRNA degradation, translational inhibition, and mRNA stabilization. They are involved in several cellular processes from development to homeostasis, and their deregulation is implicated in several diseases, including cancer. Since miRNA lacks the polyA tail, the standard single cell RNAseq protocols do not capture miRNAs, thus severely limiting our understanding of miRNA functions at cellular resolution. To overcome this limitation, we develop a novel machine learning method to infer the miRNA activity in a sample given its RNAseq profile. Methods: We develop a model using XGBoost, to predict miRNA profile in a sample from its global mRNA profile. We train and test the model using cross validation in the CCLE collection, as well as a number of healthy and cancer human tissue data obtained from GTEx and TCGA. We quantify the method's performance as the correlation between actual and predicted miRNA expression values across the test samples. We validate our model in multiple single cell datasets where miRNA and mRNA profiles are available for the same cell types by assessing the model's ability to identify cell type specific miRNAs based on a model trained on independent bulk datasets. Results: First, we show in CCLE collection, and multiple TCGA tissues, that a model based on all genes was far more accurate than the model based only on known targets. Our mean cross validation model accuracy across 10 tissues having greater than 100 paired miRNA and mRNA samples (in terms of Spearman correlation between predicted and actual expression of a miRNA) is 0.45 (min 0.39 in Pancreas to a max of 0.51 in Brain). In comparison to the normal tissues in GTEx, in the malignant counterpart in TCGA, due to greater heterogeneity, therefore greater variability in gene expression, our model performs significantly better (average cross validation accuracy improvement of 0.19). We have validated our model in independent single cell data. Using a model trained in bulk tissue data, we predict microRNA expression levels in a single cell based on the single cell RNA and compare our predicted fold difference by a miRNA's expression between two cell types with the actual fold difference. We quantify the prediction accuracy as the correlation between the predicted and actual fold differences across all miRNAs. In a total of 4 cell type pair comparisons (different sets of kidney, brain, breast, and skin), our model achieves an average accuracy of 0.81 (ranging from 0.73 to 0.89), thus strongly validating our model. Our next step is to apply our model to study miRNA activities during T cell development, Pancreatic Ductal Adenocarcinoma, and Glioblastoma, in collaboration with experimentalists. Conclusions: Our method addresses a major bottleneck in studying miRNA activities at a cellular resolution and can be applied to any scRNA data to infer miRNA activity. Citation Format: Gulden Olgun, Vishaka Gopalan, Sridhar Hannenhalli. Quantifying miRNA activity in single cell clusters [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 198.

Determination of the Bi209(n,γ)Bi210g cross section using the NICE detector

The capture cross section of $^{209}$Bi(n,$\gamma$)$^{210g}$Bi was measured at different astrophysically energies including thermal capture cross section (25 meV), resonance integral, and the Maxwellian averaged cross section at a thermal energy of $kT$ = 30 keV. The partial capture cross section ($\sigma_g$) was determined using the activation technique and by measuring the $^{210}$Po activity. The newly developed and tested NICE detector setup was used to measure the $\alpha$-activity of the $^{210}$Po. Using this setup the thermal and resonance integral cross sections were determined to be $16.2\;\pm\;0.97$~mb and $89.81\;\pm\;8.0$~mb, respectively. And the Maxwellian average cross section was measured to be $2.01\;\pm\;0.38$~mb.

Photoneutron reactions on 93Nb at Eγmax=33–93 MeV

The bremsstrahlung flux-averaged cross-sections for the photoneutron reactions 93Nb(γ,xn; x=1–5)(93-x)m,gNb were measured in the range of boundary energies of bremsstrahlung γ-quanta Eγmax=33–93 MeV. The isomeric ratios of the average cross-sections of the products of the reactions 93Nb (γ,4n)89m,gNb and 93Nb(γ,5n)88m,gNb were determined in the energy ranges Eγmax = 50–93 and 70–93 MeV, respectively. The experiments were carried out on the beam of the linear electron accelerator LUE-40 of the National Science Center Kharkov Institute of Physics and Technology using the method of induced activity. Calculations of the cross-sections, average cross-sections, and isomeric ratios of the reaction products were performed using the TALYS 1.9 code and the GEANT4.9.2 code. The experimental average cross-sections for the reactions (γ,n), (γ,2n), and (γ,4n) are in good agreement with theory, while in the case of reactions (γ,3n) and (γ,5n) some discrepancies are observed. The results obtained for the reactions (γ,n), (γ,3n), and (γ,4n) are in satisfactory agreement with the known literature data. The average cross sections for the reactions (γ,2n) and (γ,5n) and the isomeric ratios of the reaction products 93Nb(γ,5n)88m,gNb were measured for the first time.

Selection rules for the S-Matrix bootstrap

We examine the space of allowed S-matrices on the Adler zeros' plane using the recently resurrected (numerical) S-matrix bootstrap program for pion scattering. Two physical quantities, an averaged total scattering cross-section, and an averaged entanglement power for the boundary S-matrices, are studied. Emerging linearity in the leading Regge trajectory is correlated with a reduction in both these quantities. We identify two potentially viable regions where the S-matrices give decent agreement with low energy S- and P-wave scattering lengths and have leading Regge trajectory compatible with experiments. We also study the line of minimum averaged total cross section in the Adler zeros' plane. The Lovelace-Shapiro model, which was a precursor to modern string theory, is given by a straight line in the Adler zeros' plane and, quite remarkably, we find that this line intersects the space of allowed S-matrices near both these regions.

Influence of halo single-neutron transfer on near barrier ^{15}C + ^{12}C total fusion

A recent comparison of the average fusion cross section, leftlangle sigma _mathrm {F}rightrangle , for energies just above the Coulomb barrier for the ^{12-15}C + ^{12}C systems found that the behaviour as a function of projectile neutron excess could not be satisfactorily explained by static barrier penetration model calculations and suggested that the neutron dynamics plays an important rôle. In this work we demonstrate that the (^{15}C,^{14}C) single neutron transfer has a significant influence on the above barrier ^{15}C + ^{12}C total fusion, although not quite in the way expected since it leads to a reduction in the cross section, contrary to the trend in the measured leftlangle sigma _mathrm {F}rightrangle . However, this result underlines the danger of ignoring the effect of neutron transfer reactions on fusion in systems involving neutron halo nuclei.

Disease detection in rice leaves using transfer learning techniques

Recent development in deep learning techniques have had a massive impact in the field of agricultural disease detection. The negative impact of pest and bacterial diseases to rice plants are well known, and for regions where rice is staple, this is issue carries a lot of weight. This work proposes a high accuracy, transfer learned model that can provide a mobile solution for farmers and agricultural organizations to detect rice leaf diseases at hand. This study also utilizes a generative adversarial network to balance the number of disease samples. We compare our model to other transfer learning architectures as well. The presented model tested on a GAN augmented dataset, achieves an average cross validation accuracy of 98.79% outperforming paradigm classification architectures. The model is also compared on 3 different datasets, without the GAN augmentation, establishing benchmark performance of 98.38% average accuracy.

Investigation of Flux-Weighted Average Cross Sections for Reactions on $${}^{{93}}$$Nb with Bremsstrahlung of LUE-75

Investigation of Flux-Weighted Average Cross Sections for Reactions on $${}^{{93}}$$Nb with Bremsstrahlung of LUE-75

Robust Performance of Potentially Functional SNPs in Machine Learning Models for the Prediction of Atorvastatin-Induced Myalgia

Background:Statins can cause muscle symptoms resulting in poor adherence to therapy and increased cardiovascular risk. We hypothesize that combinations of potentially functional SNPs (pfSNPs), rather than individual SNPs, better predict myalgia in patients on atorvastatin. This study assesses the value of potentially functional single nucleotide polymorphisms (pfSNPs) and employs six machine learning algorithms to identify the combination of SNPs that best predict myalgia. Methods: Whole genome sequencing of 183 Chinese, Malay and Indian patients from Singapore was conducted to identify genetic variants associated with atorvastatin induced myalgia. To adjust for confounding factors, demographic and clinical characteristics were also examined for their association with myalgia. The top factor, sex, was then used as a covariate in the whole genome association analyses. Variants that were highly associated with myalgia from this and previous studies were extracted, assessed for potential functionality (pfSNPs) and incorporated into six machine learning models. Predictive performance of a combination of different models and inputs were compared using the average cross validation area under ROC curve (AUC). The minimum combination of SNPs to achieve maximum sensitivity and specificity as determined by AUC, that predict atorvastatin-induced myalgia in most, if not all the six machine learning models was determined. Results: Through whole genome association analyses using sex as a covariate, a larger proportion of pfSNPs compared to non-pf SNPs were found to be highly associated with myalgia. Although none of the individual SNPs achieved genome wide significance in univariate analyses, machine learning models identified a combination of 15 SNPs that predict myalgia with good predictive performance (AUC >0.9). SNPs within genes identified in this study significantly outperformed SNPs within genes previously reported to be associated with myalgia. pfSNPs were found to be more robust in predicting myalgia, outperforming non-pf SNPs in the majority of machine learning models tested. Conclusion: Combinations of pfSNPs that were consistently identified by different machine learning models to have high predictive performance have good potential to be clinically useful for predicting atorvastatin-induced myalgia once validated against an independent cohort of patients.

The flux weighted cross sections of 179Hf(γ,γ’)179mHf and natHf(γ,x)179mHf reactions at 8 MeV and 15 MeV bremsstrahlung end point energies

For the hafnium element, the cross sections of 179Hf(γ,γ′)179mHf and natHf(γ,x)179mHf reactions were measured at 8 MeV and 15 MeV bremsstrahlung end point energies respectively using activation method and off-line gamma-ray spectroscopy. The bremsstrahlung radiation spectra were generated by bombarding a 0.1 mm tungsten target with 8 MeV and 15 MeV electrons at 5 μA beam current and also theoretically simulated by GEANT4 computer code. The flux weighted average cross sections of 179Hf (γ,γ′)179mHf and natHf(γ,x)179mHf reactions were measured by using 115In(γ,γ’)115mIn and 197Au(γ,n)196Au as flux monitor reactions at 8 MeV and 15 MeV bremsstrahlung radiation respectively. The measured cross sections are found close to the corresponding theoretical cross sections estimated by TALYS 1.95 and TENDL 2019 computer codes. These hafnium cross sections at 8 MeV and 15 MeV bremsstrahlung radiation will be new additions to the EXFOR library, as so far not reported in literature.

A Comparison of Integrated Filtering and Prediction Methods for Smart Grids

The intelligent use of green and renewable energies requires reliable and preferably anticipated information regarding their availability and the behavior of meteorological variables in a scenario of natural intermittency. Examples of this are the smart grids, which can incorporate, among others, a charging system for electric vehicles and modern and predictive management techniques. However, some issues associated with such procedures are data captured by sensors and transducers with noise in their signals and low information repeatability under the same reading conditions. To tackle such problems, numerous filtering and data fitting techniques and various prediction methods have been developed, but an appropriate selection can be cumbersome. Also, some filtering techniques, such as RANdom SAmple Consensus (RANSAC) appear not to have been used in prediction scenarios for smart grids, to the authors’ knowledge. In this regard, this paper aims to present a comparison in terms of average error, determination coefficient, and cross validation that can be expected under prediction schemes as Multiple Linear Regression, Vector Support Machines and a Multilayer Perceptron Regression Neural Network (MLPRNN), with filtering/scaling methods such as Maximum and Minimum, L2 Norm, Standard Scale, and RANSAC. Cross validation allows to flag problems like overfitting or selection bias, and this comparison is another novelty for smart grid scenarios, to the authors’ knowledge. Although many combinations were analyzed, RANSAC, with L2 Norm filtering and an MLPRNN for prediction, generate the best results. RANSAC algorithm with L2 Norm is a novelty for filtering and predicting in smart grids, and through an MLPRNN, the R2 error can be reduced to 0.8843, the MSE to 0.8960, and the cross validation accuracy can be increased to 0.44 (±0.2).

The Scattering Properties of Gaseous Parahydrogen

In this work, the scattering properties of parahydrogen gas (p-H2) are studied in the temperature-range (1–50) K. These properties are investigated within the framework of the Lippmann–Schwinger formalism for p-H2. This formalism is used to calculate the phase shifts. The calculations are undertaken using Silvera-Goldman potential. The scattering properties of the system—the total, viscosity, and average cross sections—are then computed using these phase shifts. The s-wave scattering length for p-H2 molecules $$a_{o}$$ is determined as $$a_{o}$$ = 7.01 A. The cross sections exhibited the Ramsauer–Townsend effect and resonance peaks. The lifetime of the quasi-bound states for $$\ell - wave$$ $$\left( {\ell = {2,}\,{\text{4, and 6}}} \right)$$ is also computed.

A chemically etched corrugated feedhorn array for D-band CMB observations

We present the design, manufacturing, and testing of a 37-element array of corrugated feedhorns for Cosmic Microwave Background CMB) measurements between 140 and 170 GHz. The array was designed to be coupled to Kinetic Inductance Detector arrays, either directly (for total power measurements) or through an orthomode transducer (for polarization measurements). We manufactured the array in platelets by chemically etching aluminum plates of 0.3 mm and 0.4 mm thickness. The process is fast, low-cost, scalable, and yields high-performance antennas compared to other techniques in the same frequency range. Room temperature electromagnetic measurements show excellent repeatability with an average cross polarization level about − 20 dB, return loss about − 25 dB, first sidelobes below − 25 dB and far sidelobes below − 35 dB. Our results qualify this process as a valid candidate for state-of-the-art CMB experiments, where large detector arrays with high sensitivity and polarization purity are of paramount importance in the quest for the discovery of CMB polarization B-modes.

Function of Hemoglobin-Based Oxygen Carriers: Determination of Methemoglobin Content by Spectral Extinction Measurements.

Suspensions of hemoglobin microparticles (HbMPs) are promising tools as oxygen therapeutics. For the approval of clinical studies extensive characterization of these HbMPs with a size of about 750 nm is required regarding physical properties, function, pharmaco-kinetics and toxicology. The standard absorbance measurements in blood gas analyzers require dissolution of red blood cells which does not work for HbMP. Therefore, we have developed a robust and rapid optical method for the quality and functionality control of HbMPs. It allows simultaneous determination of the portion of the two states of hemoglobin oxygenated hemoglobin (oxyHb) and deoxygenated hemoglobin (deoxyHb) as well as the content of methemoglobin (metHb). Based on the measurement of collimated transmission spectra between 300 nm and 800 nm, the average extinction cross section of HbMPs is derived. A numerical method is applied to determine the composition of the HbMPs based on their wavelength-dependent refractive index (RI), which is a superposition of the three different states of Hb. Thus, light-scattering properties, including extinction cross sections can be simulated for different compositions and sizes. By comparison to measured spectra, the relative concentrations of oxyHb, deoxyHb, metHb are accessible. For validation of the optically determined composition of the HbMPs, we used X-ray fluorescence spectrometry for the ratio of Fe(II) (oxyHb/deoxyHb) and Fe(III) (metHb). High accuracy density measurements served to access heme-free proteins, size was determined by dynamic light scattering and analytical centrifugation and the shape of the HbMPs was visualized by electron and atomic force microscopy.

Syntheses and Crystal Structures of a Series of Dysprosium–Manganese–Sodium 12-Metallacrown-4 Compounds with Halogenated Benzoate Bridging Anions

A series of heterotrimetallic dysprosium–manganese–sodium 12-metallacrown-4 compounds has been synthesized and characterized by single-crystal X-ray analysis: DyNaX4[12-MCMn(III)N(shi)-4], where X− = 2-fluorobenzoate (1), 3-fluorobenzoate (2), 4-fluorobenzoate (3), 2-chlorobenzoate (4), 4-chlorobenzoate (5), 3-bromobenzoate (6), and 2-iodobenzoate (7); MC is metallacrown; and shi3− is salicylhydroximate. Each 12-MC-4 framework is comprised of four ring MnIII ions and four shi3− ligands. The oxime oxygen atoms of the shi3− ligands produce a central cavity which then binds one DyIII ion and one Na+ ion on opposite sides. In each molecule four halogenated benzoate anions bridge between the central DyIII ion and the ring MnIII ions via the carboxylate groups. The identity of the halogenated benzoate anion has little effect on the structure of the [12-MCMn(III)N(shi)-4] framework as determined by several metrical parameters: the MC cavity radius, the average cross cavity MnIII–MnIII distance, the average cross cavity oxime oxygen-oxime oxygen distance, and the distance of the DyIII from the oxime oxygen mean plane. For 1–7 these metrical parameters are strikingly similar; thus, providing further reinforcement of one hallmark of metallacrown chemistry—the ability to substitute various components of the molecule without affecting the overall structure which can be used for the fine-tuning of molecular properties. Four 4-fluorobenzoate (4-Fben) anions bridge between the central DyIII ion and the ring MnIII ions in the metallacrown compound DyNa(4-Fben)4[12-MCMn(III)N(shi)-4](H2O)4·4DMF.

A Multichannel Fusion Convolutional Neural Network Based on Scattering Mechanism for PolSAR Image Classification

Polarimetric features extracted from the polarimetric synthetic aperture radar data contain a wealth of target scattering information, but usually lead to the problems, such as network learning burden and high computational consumption. A multichannel fusion convolutional neural network based on scattering mechanisms was presented in this letter. First, the polarimetric features were divided into three categories according to their corresponding scattering mechanisms, and put into three network channels, respectively. Second, a new feature output was constructed based on the fusion of three-channel output features. Third, the four output features were cascaded through two fully connected layers and the Softmax classifier to get the classification result. Moreover, a new loss function was defined, combining cross entropy and average cross entropy to prevent network overfitting. Experimental results on airborne synthetic aperture radar (AIRSAR) and GF-3 data set verified the effectiveness of the proposed method in the aspect of classification accuracy and small sample.

Multi-Features Capacitive Hand Gesture Recognition Sensor: A Machine Learning Approach

Gesture recognition technology enables machines to understand human gestures. The technology is considered as a key enabler for gaming and virtual reality applications. In this paper, we propose an effective, low-cost capacitive sensor device to recognize hand gestures. In particular, we designed a prototype of a wearable capacitive sensor unit to capture the capacitance values from the electrodes placed on finger phalanges. The sensor captures finger capacitance values. Each gesture has specific finger capacitance values. We applied a running median filter to the output of the sensor and extracted 15 features for gesture classification training and testing tasks. Subsequently, various analyses were performed to provide more insights into the sensing data. We applied and compared two machine learning algorithms: Error Correction Output Code Support Vector Machines (ECOC-SVM) and ${K}$ -Nearest Neighbour (KNN) classifiers. The training and testing recognition rates were observed for both intra-participant and inter-participant data sets. Further, we introduced a feature compression approach derived from correlation analysis to reduce the complexity of the machine learning algorithms. Using cross validation, we achieved a classification rate of approximately 99% for intra-participant data. We achieved a lower recognition rate of 97% (average cross validation testing) for compressed feature data set using both machine learning approaches. For the inter-participant data, the recognition rate was 99% (normalized feature data) using KNN and 97% using ECOC-SVM. The research findings show that our recognition system is competitive and has an immense potential for further study.

RESONANCE EVALUATION OF 239Pu IN THE RESONANCE REGION UP TO 4 keV

Resolved resonance evaluation of the 239Pu cross section in the energy range up 4 keV has been carried out with the SAMMY computer code. Existing resolved resonance evaluation such as ENDF/B-VIII.0 and JENDL4 data libraries for 239Pu is limited to 2.5 keV whereas above this energy the unresolved resonance methodology is used. High resolution transmission and fission data taken at the Oak Ridge Electron Linear Accelerator (ORELA) permitted extending the resonance region up to 4 keV. The thermal and average fission cross section values calculated with the resonance parameters derived in the evaluation fall close to that indicated in the suggested IAEA fission standards.

Cross domain meta-network for sketch face recognition

Because of the large modal difference between sketch image and optical image, and the problem that traditional deep learning methods are easy to overfit in the case of a small amount of training data, the Cross Domain Meta-Network for sketch face recognition method is proposed. This method first designs a meta-learning training strategy to solve the small sample problem, and then proposes entropy average loss and cross domain adaptive loss to reduce the modal difference between the sketch domain and the optical domain. The experimental results on UoM-SGFS and PRIP-VSGC sketch face data sets show that this method and other sketch face recognition methods.

On-chip reconfigurable mode converter based on cross-connected subwavelength Y-junctions

A novel power-efficient reconfigurable mode converter is proposed and experimentally demonstrated based on cross-connected symmetric Y-junctions assisted by thermo-optic phase shifters on a silicon-on-insulator platform. Instead of using conventional Y-junctions, subwavelength symmetric Y-junctions are utilized to enhance the mode splitting ability. The reconfigurable functionality can be realized by controlling the induced phase differences. Benefited from the cross-connected scheme, the number of heating electrodes can be effectively reduced, while the performance of the device is maintained. With only one-step etching, our fabricated device shows the average insertion losses and cross talks are less than 2.45 and − 16.6 dB , respectively, measured with conversions between two arbitrary compositions of the first four TE modes over an observable 60 nm bandwidth. The converter is switchable and CMOS-compatible, and could be extended for more modes; hence, it can be potentially deployed for advanced and flexible mode multiplexing optical networks-on-chip.

SK-FMYOLOV3: A Novel Detection Method for Urine Test Strips

To accurately detect small defects in urine test strips, the SK-FMYOLOV3 defect detection algorithm is proposed. First, the prediction box clustering algorithm of YOLOV3 is improved. The fuzzy C-means clustering algorithm is used to generate the initial clustering centers, and then, the clustering center is passed to the K-means algorithm to cluster the prediction boxes. To better detect smaller defects, the YOLOV3 feature map fusion is increased from the original three-scale prediction to a four-scale prediction. At the same time, 23 convolutional layers of size 3 × 3 in the YOLOV3 network are replaced with SkNet structures, so that different feature maps can independently select different convolution kernels for training, improving the accuracy of defect classification. We collected and enhanced urine test strip images in industrial production and labeled the small defects in the images. A total of 11634 image sets were used for training and testing. The experimental results show that the algorithm can obtain an anchor frame with an average cross ratio of 86.57, while the accuracy rate and recall rate of nonconforming products are 96.8 and 94.5, respectively. The algorithm can also accurately identify the category of defects in nonconforming products.