EM-X-DL: Efficient Cross-device Deep Learning Side-channel Attack With Noisy EM Signatures

This work presents a Cross-device Deep-Learning based Electromagnetic (EM-X-DL) side-channel analysis (SCA) on AES-128, in the presence of a significantly lower signal-to-noise ratio (SNR) compared to previous works. Using a novel algorithm to intelligently select multiple training devices and proper choice of hyperparameters, the proposed 256-class deep neural network (DNN) can be trained efficiently utilizing pre-processing techniques like PCA, LDA, and FFT on measurements from the target encryption engine running on an 8-bit Atmel microcontroller. In this way, EM-X-DL achieves >90% single-trace attack accuracy. Finally, an efficient end-to-end SCA leakage detection and attack framework using EM-X-DL demonstrates high confidence of an attacker with <20 averaged EM traces.

Large-scale protein-protein post-translational modification extraction with distant supervision and confidence calibrated BioBERT

Motivation Protein-protein interactions (PPIs) are critical to normal cellular function and are related to many disease pathways. A range of protein functions are mediated and regulated by protein interactions through post-translational modifications (PTM). However, only 4% of PPIs are annotated with PTMs in biological knowledge databases such as IntAct, mainly performed through manual curation, which is neither time- nor cost-effective. Here we aim to facilitate annotation by extracting PPIs along with their pairwise PTM from the literature by using distantly supervised training data using deep learning to aid human curation. Method We use the IntAct PPI database to create a distant supervised dataset annotated with interacting protein pairs, their corresponding PTM type, and associated abstracts from the PubMed database. We train an ensemble of BioBERT models—dubbed PPI-BioBERT-x10—to improve confidence calibration. We extend the use of ensemble average confidence approach with confidence variation to counteract the effects of class imbalance to extract high confidence predictions. Results and conclusion The PPI-BioBERT-x10 model evaluated on the test set resulted in a modest F1-micro 41.3 (P =5 8.1, R = 32.1). However, by combining high confidence and low variation to identify high quality predictions, tuning the predictions for precision, we retained 19% of the test predictions with 100% precision. We evaluated PPI-BioBERT-x10 on 18 million PubMed abstracts and extracted 1.6 million (546507 unique PTM-PPI triplets) PTM-PPI predictions, and filter $$\approx 5700$$ ≈ 5700 (4584 unique) high confidence predictions. Of the 5700, human evaluation on a small randomly sampled subset shows that the precision drops to 33.7% despite confidence calibration and highlights the challenges of generalisability beyond the test set even with confidence calibration. We circumvent the problem by only including predictions associated with multiple papers, improving the precision to 58.8%. In this work, we highlight the benefits and challenges of deep learning-based text mining in practice, and the need for increased emphasis on confidence calibration to facilitate human curation efforts.

An Analysis of the Accuracy of Photo-Based Plant Identification Applications on Fifty-Five Tree Species

Background: With the creation of photo-based plant identification applications (apps), the ability to attain basic identifications of plants in the field is seemingly available to anyone who has access to a smartphone. The use of such apps as an educational tool for students and as a major identification resource for some community science projects calls into question the accuracy of the identifications they provide. We created a study based on the context of local tree species in order to offer an informed response to students asking for guidance when choosing a tool for their support in classes. Methods: This study tested 6 mobile plant identification apps on a set of 440 photographs representing the leaves and bark of 55 tree species common to the state of New Jersey (USA). Results: Of the 6 apps tested, PictureThis was the most accurate, followed by iNaturalist, with PlantSnap failing to offer consistently accurate identifications. Overall, these apps are much more accurate in identifying leaf photos as compared to bark photos, and while these apps offer accurate identifications to the genus-level, there seems to be little accuracy in successfully identifying photos to the species-level. Conclusions: While these apps cannot replace traditional field identification, they can be used with high confidence as a tool to assist inexperienced or unsure arborists, foresters, or ecologists by helping to refine the pool of possible species for further identification.

Real-time diagnostic accuracy of blue light imaging, linked color imaging and white-light endoscopy for colorectal polyp characterization

Background and study aims Fujifilm has developed a novel ELUXEO 7000 endoscope system that employs light-emitting diodes (LEDs) at four different wavelengths as light sources that enable blue light imaging (BLI), linked color imaging (LCI), and high-definition white-light endoscopy (HD-WLE). The aim of this study was to address the diagnostic accuracy of real-time polyp characterization using BLI, LCI and HD-WLE (ELUXEO 7000 endoscopy system). Patients methods This is a prespecified post-hoc analysis of a prospective study in which 22 experienced endoscopists (> 2,000 colonoscopies) from eight international centers participated. Using a combination of BLI, LCI, and HD-WLE, lesions were endoscopically characterized including a high- or low-confidence statement. Per protocol, digital images were created from all three imaging modalities. Histopathology was the reference standard. Endoscopists were familiar with polyp characterization, but did not take dedicated training for purposes of this study. Results Overall, 341 lesions were detected in 332 patients. Of the lesions, 269 histologically confirmed polyps with an optical diagnosis were included for analysis (165 adenomas, 27 sessile serrated lesions, and 77 hyperplastic polyps). Overall, polyp characterization was performed with high confidence in 82.9 %. The overall accuracy for polyp characterization was 75.1 % (95 % confidence interval [CI] 69.5–80.1 %), compared with an accuracy of 78.0 % (95 % CI 72.0–83.2 %) for high confidence assignments. The accuracy for endoscopic characterization for diminutive polyps was 74.7 % (95 %CI 68.4–80.3 %), compared with an accuracy of 78.2 % (95 % CI 71.4–84.0 %) for high-confidence assignments. Conclusions The diagnostic accuracy of BLI, LCI, and HD-WLE by experienced endoscopist for real-time polyp characterization seems limited (NCT03344289).

FlyPhoneDB: An integrated web-based resource for cell-cell communication prediction in Drosophila

Multicellular organisms rely on cell-cell communication to exchange information necessary for developmental processes and metabolic homeostasis. Cell-cell communication pathways can be inferred from transcriptomic datasets based on ligand-receptor (L-R) expression. Recently, data generated from single cell RNA sequencing (scRNA-seq) have enabled L-R interaction predictions at an unprecedented resolution. While computational methods are available to infer cell-cell communication in vertebrates such a tool does not yet exist for Drosophila. Here, we generated a high confidence list of L-R pairs for the major fly signaling pathways and developed FlyPhoneDB, a quantification algorithm that calculates interaction scores to predict L-R interactions between cells. At the FlyPhoneDB user interface, results are presented in a variety of tabular and graphical formats to facilitate biological interpretation. To demonstrate that FlyPhoneDB can effectively identify active ligands and receptors to uncover cell-cell communication events, we applied FlyPhoneDB to Drosophila scRNA-seq data sets from adult midgut, abdomen, and blood, and demonstrate that FlyPhoneDB can readily identify previously characterized cell-cell communication pathways. Altogether, FlyPhoneDB is an easy-to-use framework that can be used to predict cell-cell communication between cell types from scRNA-seq data in Drosophila.

Ataxin-2, Twenty-four and Dicer-2 are components of a non-canonical cytoplasmic polyadenylation complex

Cytoplasmic polyadenylation is a mechanism to promote mRNA translation in a wide variety of biological contexts. A canonical complex centered around the conserved RNA-binding protein family CPEB has been shown to be responsible for this process. We have previously reported evidence for an alternative non-canonical, CPEB-independent complex in Drosophila, of which the RNA-interference factor Dicer-2 is a component. Here, we investigate Dicer-2 mRNA targets and protein co-factors in cytoplasmic polyadenylation. Using RIP-Seq analysis we identify hundreds of novel Dicer-2 target transcripts, ~50% of which were previously found as targets of the cytoplasmic poly(A) polymerase Wispy, suggesting widespread roles of Dicer-2 in cytoplasmic polyadenylation. Large-scale immunoprecipitation revealed Ataxin-2 and Twenty-four among the high-confidence interactors of Dicer-2. Functional analysis indicate that both factors form an RNA-independent complex with Dicer-2, and are required for cytoplasmic polyadenylation of Dicer-2 targets. Our results reveal the composition of a novel cytoplasmic polyadenylation complex that operates during Drosophila early embryogenesis.

Exome Sequencing Reveals Novel Variants and Expands the Genetic Landscape for Congenital Microcephaly

Congenital microcephaly causes smaller than average head circumference relative to age, sex and ethnicity and is most usually associated with a variety of neurodevelopmental disorders. The underlying etiology is highly heterogeneous and can be either environmental or genetic. Disruption of any one of multiple biological processes, such as those underlying neurogenesis, cell cycle and division, DNA repair or transcription regulation, can result in microcephaly. This etiological heterogeneity manifests in a clinical variability and presents a major diagnostic and therapeutic challenge, leaving an unacceptably large proportion of over half of microcephaly patients without molecular diagnosis. To elucidate the clinical and genetic landscapes of congenital microcephaly, we sequenced the exomes of 191 clinically diagnosed patients with microcephaly as one of the features. We established a molecular basis for microcephaly in 71 patients (37%), and detected novel variants in five high confidence candidate genes previously unassociated with this condition. We report a large number of patients with mutations in tubulin-related genes in our cohort as well as higher incidence of pathogenic mutations in MCPH genes. Our study expands the phenotypic and genetic landscape of microcephaly, facilitating differential clinical diagnoses for disorders associated with most commonly disrupted genes in our cohort.

Assessing the benefits of Imaging Infrared Radiometer observations to the CALIOP version 4 cloud and aerosol discrimination algorithm

The features detected in monolayer atmospheric columns sounded by the Cloud and Aerosol Lidar with Orthogonal Polarization (CALIOP) and classified as cloud or aerosol layers by the CALIOP version 4 (V4) cloud and aerosol discrimination (CAD) algorithm are reassessed using perfectly collocated brightness temperatures measured by the Imaging Infrared Radiometer (IIR) onboard the same satellite. Using the IIR’s three wavelength measurements of layers that are confidently classified by the CALIOP CAD algorithm, we calculate two-dimensional (2-D) probability distribution functions (PDFs) of IIR brightness temperature differences (BTDs) for different cloud and aerosol types. We then compare these PDFs with 1-D radiative transfer simulations for ice and water clouds and dust and marine aerosols. Using these IIR 2-D BTD signature PDFs, we develop and deploy a new IIR-based CAD algorithm and compare the classifications obtained to the results reported by the CALIOP-only V4 CAD algorithm. IIR observations are shown to be able to identify clouds with a good accuracy. The IIR cloud identifications agree very well with layers classified as confident clouds by the V4 CAD algorithm (88 %). More importantly, simultaneous use of IIR information reduces the ambiguity in a notable fraction of "not confident" V4 cloud classifications. 28 % and 14 % of the ambiguous V4 cloud classifications are confirmed thanks to the IIR observations in the tropics and in the midlatitudes respectively. IIR observations are of relatively little help in deriving high confidence classifications for most aerosols, as the low altitudes and small optical depths of aerosol layers yield IIR signatures that are similar to those from clear skies. However, misclassifications of aerosol layers, such as dense dust or elevated smoke layers, by the V4 CAD algorithm can be corrected to cloud layer classification by including IIR information. 10 %, 16 %, and 6 % of the ambiguous V4 dust, polluted dust, and tropospheric elevated smoke respectively are found to be misclassified cloud layers by the IIR measurements.