Many communications and sensing applications hinge on the detection of a signal in a noisy, interference-heavy environment. Signal processing theory yields techniques such as the generalized ...likelihood ratio test (GLRT) to perform detection when the received samples correspond to a linear observation model. Numerous practical applications exist, however, where the received signal has passed through a nonlinearity, causing significant performance degradation of the GLRT. In this work, we propose prepending the GLRT detector with a neural network classifier capable of identifying the particular nonlinear time samples in a received signal. We show that pre-processing received nonlinear signals using our trained classifier to eliminate excessively nonlinear samples (i) improves the detection performance of the GLRT on nonlinear signals and (ii) retains the theoretical guarantees provided by the GLRT on linear observation models for accurate signal detection.
Deep learning-based automatic modulation classification (AMC) models are susceptible to adversarial attacks. Such attacks inject specifically crafted wireless interference into transmitted signals to ...induce erroneous classification predictions. Furthermore, adversarial interference is transferable in black box environments, allowing an adversary to attack multiple deep learning models with a single perturbation crafted for a particular classification model. In this work, we propose a novel wireless receiver architecture to mitigate the effects of adversarial interference in various black box attack environments. We begin by evaluating the architecture uncertainty environment, where we show that adversarial attacks crafted to fool specific AMC DL architectures are not directly transferable to different DL architectures. Next, we consider the domain uncertainty environment, where we show that adversarial attacks crafted on time domain and frequency domain features to not directly transfer to the altering domain. Using these insights, we develop our Assorted Deep Ensemble (ADE) defense, which is an ensemble of deep learning architectures trained on time and frequency domain representations of received signals. Through evaluation on two wireless signal datasets under different sources of uncertainty, we demonstrate that our ADE obtains substantial improvements in AMC classification performance compared with baseline defenses across different adversarial attacks and potencies.
Aberrant activation of the SRC family kinase hematopoietic cell kinase (HCK) triggers hematological malignancies as a tumor cell-intrinsic oncogene. Here we find that high HCK levels correlate with ...reduced survival of colorectal cancer patients. Likewise, increased Hck activity in mice promotes the growth of endogenous colonic malignancies and of human colorectal cancer cell xenografts. Furthermore, tumor-associated macrophages of the corresponding tumors show a pronounced alternatively activated endotype, which occurs independently of mature lymphocytes or of Stat6-dependent Th2 cytokine signaling. Accordingly, pharmacological inhibition or genetic reduction of Hck activity suppresses alternative activation of tumor-associated macrophages and the growth of colon cancer xenografts. Thus, Hck may serve as a promising therapeutic target for solid malignancies.
Display omitted
•Abundant HCK in tumor leukocytes of human colon cancer correlates with poor survival•Excessive myeloid HCK activity results in alternative macrophage polarization•Myeloid HCK promotes colon tumorigenesis associated with increased Stat3 activity•Ablation of HCK or its therapeutic inhibition limits colon cancer xenograft growth
Poh et al. show that myeloid-specific Hck activity regulates tumor-associated macrophage polarization, the accumulation of IL-6/IL-11 family cytokines and colon cancer growth. Inhibition of Hck activity reduces tumor burden in mice. In human colorectal cancer, high HCK expression correlates with poor patient outcome.
The transcriptome of the developing starchy endosperm of hexaploid wheat (Triticum aestivum) was determined using RNASeq isolated at five stages during grain fill. This resource represents an ...excellent way to identify candidate genes responsible for the starchy endosperm cell wall, which is dominated by arabinoxylan (AX), accounting for 70% of the cell wall polysaccharides, with 20% (1,3; 1,4)-β-D-glucan, 7% glucomannan, and 4% cellulose. A complete inventory of transcripts of 124 glycosyltransferase (GT) and 72 glycosylhydrolase (GH) genes associated with cell walls is presented. The most highly expressed GT transcript (excluding those known to be involved in starch synthesis) was a GT47 family transcript similar to Arabidopsis (Arabidopsis thaliana) IRX10 involved in xylan extension, and the second most abundant was a GT61. Profiles for GT43 IRX9 and IRX14 putative orthologs were consistent with roles in AX synthesis. Low abundances were found for transcripts from genes in the acyl-coA transferase BAHD family, for which a role in AX feruloylation has been postulated. The relative expression of these was much greater in whole grain compared with starchy endosperm, correlating with the levels of bound ferulate. Transcripts associated with callose (GSL), cellulose (CESA), pectin (GAUT), and glucomannan (CSLA) synthesis were also abundant in starchy endosperm, while the corresponding cell wall polysaccharides were confirmed as low abundance (glucomannan and callose) or undetectable (pectin) in these samples. Abundant transcripts from GH families associated with the hydrolysis of these polysaccharides were also present, suggesting that they may be rapidly turned over.Abundant transcripts in the GT31 family may be responsible for the addition of Gal residues to arabinogalactan peptide.
People with symptoms of depression show impairments in decision-making. One explanation is that they have difficulty maintaining rich representations of the task environment. We test this hypothesis ...in the context of exploratory choice. We analyze depressive and non-depressive participants’ exploration strategies by comparing their choices to two computational models: (1) an “Ideal Actor” model that reflectively updates beliefs and plans ahead, employing a rich representation of the environment and (2) a “Naïve Reinforcement Learning” (RL) model that updates beliefs reflexively utilizing a minimal task representation. Relative to non-depressive participants, we find that depressive participants’ choices are better described by the simple RL model. Further, depressive participants were more exploratory than non-depressives in their decision-making. Depressive symptoms appear to influence basic mechanisms supporting choice behavior by reducing use of rich task representations and hindering performance during exploratory decision-making.
Recent work has advocated for the use of deep learning to perform power allocation in the downlink of massive MIMO (maMIMO) networks. Yet, such deep learning models are vulnerable to adversarial ...attacks. In the context of maMIMO power allocation, adversarial attacks refer to the injection of subtle perturbations into the deep learning model's input, during inference (i.e., the adversarial perturbation is injected into inputs during deployment after the model has been trained) that are specifically crafted to force the trained regression model to output an infeasible power allocation solution. In this work, we develop an autoencoder-based mitigation technique, which allows deep learning-based power allocation models to operate in the presence of adversaries without requiring retraining. Specifically, we develop a denoising autoencoder (DAE), which learns a mapping between potentially perturbed data and its corresponding unperturbed input. We test our defense across multiple attacks and in multiple threat models and demonstrate its ability to (i) mitigate the effects of adversarial attacks on power allocation networks using two common precoding schemes, (ii) outperform previously proposed benchmarks for mitigating regression-based adversarial attacks on maMIMO networks, (iii) retain accurate performance in the absence of an attack, and (iv) operate with low computational overhead.
The Deepwater Horizon (DWH) oil spill caused an estimated 100,000 bird mortalities. However, mortality estimates are often based on the number of visibly oiled birds and likely underestimate the true ...damage to avian populations as they do not include toxic effects from crude oil ingestion. Elevated susceptibility to disease has been postulated to be a significant barrier to recovery for birds that have ingested crude oil. Effective defense against pathogens involves integration of physiological and behavioral traits, which are regulated in-part by cytokine signaling pathways. In this study, we tested whether crude oil ingestion altered behavioral and physiological aspects of disease defense in birds. To do so, we used artificially weathered Mississippi Canyon 242 crude oil to orally dose zebra finches (Taeniopygia guttata) with 3.3 mL/kg or 10 mL/kg of crude oil or a control (peanut oil) for 14 days. We measured expression of cytokines (interleukin IL-1β, IL-6, IL-10) and proinflammatory pathways (NF-κB, COX-2) in the intestine, liver, and spleen (tissues that exhibit pathology in oil-exposed birds). We also measured heterophil:lymphocyte (H:L) ratio and complement system activity, and video-recorded birds to analyze sickness behavior. Finches that ingested crude oil exhibited tissue-specific changes in cytokine mRNA expression. Proinflammatory cytokine expression decreased in the intestine but increased in the liver and spleen. Birds exposed to crude oil had lower H:L ratios compared to the control on day 14, but there were no differences in complement activity among treatments. Additionally, birds exposed to 10 mL/kg crude oil had reduced activity, indicative of sickness behavior. Our results suggest cytokines play a role in mediating physiological and behavioral responses to crude oil ingestion. Although most avian population damage assessments focus on mortality caused by external oiling, crude oil ingestion may also indirectly affect survival by altering physiological and behavioral traits important for disease defense.
Display omitted
•Crude oil ingestion alters cytokine mRNA expression in tissues.•Oil-exposed birds had a lower heterophil:lymphocyte ratio.•Oil-exposed birds induced sickness behaviors.•Altered cytokine expression may contribute to immunotoxicity.
Crude oil ingestion impaired physiological and behavioral components of disease defense, which may be partially attributable to shifts in cytokine mRNA expression.
Abstract Background and Aims Proteomics holds promise for individualizing cancer treatment. We analyzed to what extent the proteomic landscape of human colorectal cancer (CRC) is maintained in ...established CRC cell lines and the utility of proteomics for predicting therapeutic responses. Methods Proteomic and transcriptomic analyses were performed on 44 CRC cell lines, compared against primary CRCs (n=95) and normal tissues (n=60), and integrated with genomic and drug sensitivity data. Results Cell lines mirrored the proteomic aberrations of primary tumors, in particular for intrinsic programs. Tumor relationships of protein expression with DNA copy number aberrations and signatures of post-transcriptional regulation were recapitulated in cell lines. The five proteomic subtypes previously identified in tumors were represented among cell lines. Nonetheless, systematic differences between cell line and tumor proteomes were apparent, attributable to stroma, extrinsic signaling and growth conditions. Contribution of tumor stroma obscured signatures of DNA mismatch repair identified in cell lines with a hypermutation phenotype. Global proteomic data showed improved utility for predicting both known drug-target relationships and overall drug sensitivity as compared to genomic or transcriptomic measurements. Inhibition of targetable proteins associated with drug responses further identified corresponding synergistic or antagonistic drug combinations. Our data provide evidence for CRC proteomic subtype-specific drug responses. Conclusions Proteomes of established CRC cell line are representative of primary tumors. Proteomic data tend to exhibit improved prediction of drug sensitivity as compared to genomic and transcriptomic profiles. Our integrative proteogenomic analysis highlights the potential of proteome profiling to inform personalized cancer medicine.
Next generation sequencing for oncology patient management is now routine in clinical pathology laboratories. Although wet lab, sequencing and pipeline tasks are largely automated, the analysis of ...variants for clinical reporting remains largely a manual task. The increasing volume of sequencing data and the limited availability of genetic experts to analyse and report on variants in the data is a key scalability limit for molecular diagnostics.
To determine the impact and size of the issue, we examined the longitudinally compiled genetic variants from 48,036 cancer patients over a six year period in a large cancer hospital from ten targeted cancer panel tests in germline, solid tumour and haematology contexts using hybridization capture and amplicon assays. This testing generated 24,168,398 sequenced variants of which 23,255 (8214 unique) were clinically reported.
Of the reported variants, 17,240 (74.1%) were identified in more than one assay which allowed curated variant data to be reused in later reports. The remainder, 6015 (25.9%) were not subsequently seen in later assays and did not provide any reuse benefit. The number of new variants requiring curation has significantly increased over time from 1.72 to 3.73 variants per sample (292 curated variants per month). Analysis of the 23,255 variants reported, showed 28.6% (n = 2356) were not present in common public variant resources and therefore required de novo curation. These in-house only variants were enriched for indels, tumour suppressor genes and from solid tumour assays.
This analysis highlights the significant percentage of variants not present within common public variant resources and the level of non-recurrent variants that consequently require greater curation effort. Many of these variants are unique to a single patient and unlikely to appear in other patients reflecting the personalised nature of cancer genomics. This study depicts the real-world situation for pathology laboratories faced with curating increasing numbers of low-recurrence variants while needing to expedite the process of manual variant curation. In the absence of suitably accurate automated methods, new approaches are needed to scale oncology diagnostics for future genetic testing volumes.
Automatic modulation classification (AMC) aims to improve the efficiency of crowded radio spectrums by automatically predicting the modulation constellation of wireless RF signals. Recent work has ...demonstrated the ability of deep learning to achieve robust AMC performance using raw in-phase and quadrature (IQ) time samples. Yet, deep learning models are highly susceptible to adversarial interference, which cause intelligent prediction models to misclassify received samples with high confidence. Furthermore, adversarial interference is often transferable, allowing an adversary to attack multiple deep learning models with a single perturbation crafted for a particular classification network. In this work, we present a novel receiver architecture consisting of deep learning models capable of withstanding transferable adversarial interference. Specifically, we show that adversarial attacks crafted to fool models trained on time-domain features are not easily transferable to models trained using frequency-domain features. In this capacity, we demonstrate classification performance improvements greater than 30% on recurrent neural networks (RNNs) and greater than 50% on convolutional neural networks (CNNs). We further demonstrate our frequency feature-based classification models to achieve accuracies greater than 99% in the absence of attacks.