Search

Cell type identification using single-cell RNA sequencing data is critical for understanding disease mechanisms and drug discovery. Cell clustering analysis has been widely studied in health research for rare tumor cell detection. In this study, we propose a Gaussian mixture model-based variational graph autoencoder on scRNA-seq data (scGMM-VGAE) that integrates a statistical clustering model to a deep learning algorithm to significantly improve the cell clustering performance. This model feeds a cell-cell graph adjacency matrix and a gene feature matrix into a graph variational autoencoder (VGAE) to generate latent data. These data are then used for cell clustering by the Gaussian mixture model (GMM) module. To optimize the algorithm, a designed loss function is derived by combining parameter estimates from the GMM and VGAE. We test the proposed method on four publicly available and three simulated datasets which contain many biological and technical zeros. The scGMM-VGAE outperforms four selected baseline methods on three evaluation metrics in cell clustering. By successfully incorporating GMM into deep learning VGAE on scRNA-seq data, the proposed method shows higher accuracy in cell clustering on scRNA-seq data. This improvement has a significant impact on detecting rare cell types in health research. All source codes used in this study can be found at https://github.com/ericlin1230/scGMM-VGAE .

BackgroundAlthough electronic health record systems have facilitated clinical documentation in health care, they have also introduced new challenges, such as the proliferation of redundant information through the use of copy and paste commands or templates. One approach to trimming down bloated clinical documentation and improving clinical summarization is to identify highly similar text snippets with the goal of removing such text. ObjectiveWe developed a natural language processing system for the task of assessing clinical semantic textual similarity. The system assigns scores to pairs of clinical text snippets based on their clinical semantic similarity. MethodsWe leveraged recent advances in natural language processing and graph representation learning to create a model that combines linguistic and domain knowledge information from the MedSTS data set to assess clinical semantic textual similarity. We used bidirectional encoder representation from transformers (BERT)–based models as text encoders for the sentence pairs in the data set and graph convolutional networks (GCNs) as graph encoders for corresponding concept graphs that were constructed based on the sentences. We also explored techniques, including data augmentation, ensembling, and knowledge distillation, to improve the model’s performance, as measured by the Pearson correlation coefficient (r). ResultsFine-tuning the BERT_base and ClinicalBERT models on the MedSTS data set provided a strong baseline (Pearson correlation coefficients: 0.842 and 0.848, respectively) compared to those of the previous year’s submissions. Our data augmentation techniques yielded moderate gains in performance, and adding a GCN-based graph encoder to incorporate the concept graphs also boosted performance, especially when the node features were initialized with pretrained knowledge graph embeddings of the concepts (r=0.868). As expected, ensembling improved performance, and performing multisource ensembling by using different language model variants, conducting knowledge distillation with the multisource ensemble model, and taking a final ensemble of the distilled models further improved the system’s performance (Pearson correlation coefficients: 0.875, 0.878, and 0.882, respectively). ConclusionsThis study presents a system for the MedSTS clinical semantic textual similarity benchmark task, which was created by combining BERT-based text encoders and GCN-based graph encoders in order to incorporate domain knowledge into the natural language processing pipeline. We also experimented with other techniques involving data augmentation, pretrained concept embeddings, ensembling, and knowledge distillation to further increase our system’s performance. Although the task and its benchmark data set are in the early stages of development, this study, as well as the results of the competition, demonstrates the potential of modern language model–based systems to detect redundant information in clinical notes.

A variety of genetic techniques have been devised to determine cell lineage relationships during tissue development. Some of these systems monitor cell lineages spatially and/or temporally without regard to gene expression by the cells, whereas others correlate gene expression with the lineage under study. The GAL4 Technique for Real-time and Clonal Expression (G-TRACE) system allows for rapid, fluorescent protein-based visualization of both current and past GAL4 expression patterns and is therefore amenable to genome-wide expression-based lineage screens. Here we describe the results from such a screen, performed by undergraduate students of the University of California, Los Angeles (UCLA) Undergraduate Research Consortium for Functional Genomics (URCFG) and high school summer scholars as part of a discovery-based education program. The results of the screen, which reveal novel expression-based lineage patterns within the brain, the imaginal disc epithelia, and the hematopoietic lymph gland, have been compiled into the G-TRACE Expression Database (GED), an online resource for use by the Drosophila research community. The impact of this discovery-based research experience on student learning gains was assessed independently and shown to be greater than that of similar programs conducted elsewhere. Furthermore, students participating in the URCFG showed considerably higher STEM retention rates than UCLA STEM students that did not participate in the URCFG, as well as STEM students nationwide.

Transcriptome analysis of circulating tumor cells (CTCs) provides insights into monitoring target therapeutics and underlying tumor metastasis. Here the authors present Hydro-Seq, a contamination-free high-throughput hydrodynamic scRNA-seq barcoding technique for rare CTCs.

Summary: Ibrutinib (IB) is an oral Bruton's tyrosine kinase (BTK) inhibitor that has demonstrated benefit in B cell cancers, but is associated with a dramatic increase in atrial fibrillation (AF). We employed cell-specific differentiation protocols and optical mapping to investigate the effects of IB and other tyrosine kinase inhibitors (TKIs) on the voltage and calcium transients of atrial and ventricular human pluripotent stem cell-derived cardiomyocytes (hPSC-CMs). IB demonstrated direct cell-specific effects on atrial hPSC-CMs that would be predicted to predispose to AF. Second-generation BTK inhibitors did not have the same effect. Furthermore, IB exposure was associated with differential chamber-specific regulation of a number of regulatory pathways including the receptor tyrosine kinase pathway, which may be implicated in the pathogenesis of AF. Our study is the first to demonstrate cell-type-specific toxicity in hPSC-derived atrial and ventricular cardiomyocytes, which reliably reproduces the clinical cardiotoxicity observed. : The authors employ cell-specific cardiac differentiation protocols, RNA-seq, and optical mapping to demonstrate atrial-specific toxicity of ibrutinib, a first-in-class BTK inhibitor. Other tyrosine kinase inhibitors (TKIs) with the same drug target do not affect atrial electrophysiology. Nilotinib and vandetanib, two TKIs known to be associated with QT prolongation and risk of sudden death, demonstrated ventricular-specific electrophysiologic dysregulation. Keywords: cardiac electrophysiology, tyrosine kinase inhibitors, atrial fibrillation, drug screening, optical mapping, RNA-seq

Feeding is critical for survival, and disruption in the mechanisms that govern food intake underlies disorders such as obesity and anorexia nervosa. It is important to understand both food intake and food motivation to reveal mechanisms underlying feeding disorders. Operant behavioral testing can be used to measure the motivational component to feeding, but most food intake monitoring systems do not measure operant behavior. Here, we present a new solution for monitoring both food intake and motivation in rodent home-cages: the Feeding Experimentation Device version 3 (FED3). FED3 measures food intake and operant behavior in rodent home-cages, enabling longitudinal studies of feeding behavior with minimal experimenter intervention. It has a programmable output for synchronizing behavior with optogenetic stimulation or neural recordings. Finally, FED3 design files are open-source and freely available, allowing researchers to modify FED3 to suit their needs.

Caspase-8 (CASP8) is one of the most frequently mutated genes in head and neck squamous carcinomas (HNSCCs), and CASP8 mutations are associated with poor survival. The distribution of these mutations in HNSCCs suggests that they are likely to be inactivating. Inhibition of CASP8 has been reported to sensitize cancer cells to necroptosis, a regulated cell death mechanism. Here, we show that knockdown of CASP8 renders HNSCCs susceptible to necroptosis by a second mitochondria-derived activator of caspase (SMAC) mimetic, birinapant, in combination with pan-caspase inhibitors Z-VAD-FMK or emricasan and radiation. In a syngeneic mouse model of oral cancer, birinapant, particularly when combined with radiation, delayed tumor growth and enhanced survival under CASP8 loss. Exploration of molecular underpinnings of necroptosis sensitivity confirmed that the level of functional receptor-interacting serine/threonine protein kinase 3 (RIP3) determines susceptibility to this mode of death. Although an in vitro screen revealed that low RIP3 levels rendered many HNSCC cell lines resistant to necroptosis, patient tumors maintained RIP3 expression and should therefore remain sensitive. Collectively, these results suggest that targeting the necroptosis pathway with SMAC mimetics, especially in combination with radiation, may be relevant therapeutically in HNSCC with compromised CASP8 status, provided that RIP3 function is maintained.

Abstract Since current experimental models of Atrial Fibrillation (AF) have significant limitations, we used human embryonic stem cells (hESCs) to generate an atrial-specific tissue model of AF for pharmacologic testing. We generated atrial-like cardiomyocytes (CMs) from hESCs which preferentially expressed atrial-specific genes, and had shorter action potential (AP) durations compared to ventricular-like CMs. We then generated confluent atrial-like CM sheets and interrogated them using optical mapping techniques. Atrial-like CM sheets (~1 cm in diameter) showed uniform AP propagation, and rapid re-entrant rotor patterns, as seen in AF could be induced. Anti-arrhythmic drugs were tested on single atrial-like CMs and cell sheets. Flecainide profoundly slowed upstroke velocity without affecting AP duration, leading to reduced conduction velocities (CVs), curvatures and cycle lengths of rotors, consistent with increased rotor organization and expansion. By contrast, consistent with block of rapid delayed rectifier K+ currents (Ikr) and AP prolongation in isolated atrial-like CMs, dofetilide prolonged APs and reduced cycle lengths of rotors in cell sheets without affecting CV. In conclusion, using our hESC-derived atrial CM preparations, we demonstrate that flecainide and dofetilide modulate reentrant arrhythmogenic rotor activation patterns in a manner that helps explain their efficacy in treating and preventing AF.

The novel positive-contrast magnetic resonance imaging (MRI) marker C4 consists of an aqueous solution of cobalt chloride (CoCl2) complexed with the chelator N-acetylcysteine (NAC). We evaluated whether the presence of C4 or its components would produce reactive oxygen species (ROS, including hydroxyl, peroxyl, or other reactive oxygen species) in cultured cells. Human cancer or normal cells were incubated with 1% (w/v) CoCl2·6H2O or 2% NAC or a combination of both (1% CoCl2·6H2O : 2% NAC in an aqueous solution, abbreviated as Co : NAC) in the presence or absence of H2O2. Intracellular ROS levels were measured and quantified by change in relative fluorescence units. Student’s t-tests were used. In all cell lines exposed to 1000 μM H2O2, the Co : NAC led to ≥94.7% suppression of ROS at 5 minutes and completely suppressed ROS at 60 and 90 minutes; NAC suppressed ROS by ≥76.6% at 5 minutes and by ≥94.5% at 90 minutes; and CoCl2·6H2O suppressed ROS by ≥37.2% at 30 minutes and by ≥48.6% at 90 minutes. These results demonstrate that neither Co : NAC nor its components generated ROS; rather, they suppressed ROS production in cultured cells, suggesting that C4 would not enhance ROS production in clinical use.

Following in the footsteps of the field Networked Infomechanical Systems (NIMS) currently deployed at the James San Jacinto Mountain Reserve in Idyllwild California, two smaller scale NIMS architectures have emerged. The Laboratory Systems (NIMS-LS) provides an experimental testbed for verification of actuated sensing of environmental phenomena. The NIMS-LS System also includes emulation of environmental phenomena (i.e. light distribution in ecosystems). The Rapidly Deployable (NIMS-RD) platform provides a flexible means of sampling environmental phenomena directly with minimal infrastructure, rapid deployability and minimal impact on the environment. NIMS-RD has been deployed at the James San Jacinto Mountain Reserve, Medea Creek in the Santa Monica Mountains, Merced River in Merced CA and White Mountains Research Station near Bishop CA.

Rapidly growing stress on the natural environment has created an urgent demand for new environmental exploration systems that operate remotely and autonomously. As sensor networks are applied to these critical environmental monitoring problems, a series of new requirements have emerged for extensive characterization of dynamic phenomena in three-dimensional environments. The unpredictable presence of obstacles to sensing and the inherent temporal evolution of events create limitations for optimal deployment of sensors in the environment. Networked Infomechanical Systems (NIMS) technology has been developed to introduce precise and sustainable actuation for distributed sensing. NIMS devices are suspended on cable infrastructure to provide a combination of broad range motion combined with accurate localization.

Rapidly growing stress on the natural environment has created an urgent demand for new environmental exploration systems that operate remotely and autonomously. As sensor networks are applied to these critical environmental monitoring problems, a series of new requirements have emerged for extensive characterization of dynamic phenomena in three-dimensional environments. The unpredictable presence of obstacles to sensing and the inherent temporal evolution of events create limitations for optimal deployment of sensors in the environment.

Abstract Machine-learned potentials (MLPs) have exhibited remarkable accuracy, yet the lack of general-purpose MLPs for a broad spectrum of elements and their alloys limits their applicability. Here, we present a promising approach for constructing a unified general-purpose MLP for numerous elements, demonstrated through a model (UNEP-v1) for 16 elemental metals and their alloys. To achieve a complete representation of the chemical space, we show, via principal component analysis and diverse test datasets, that employing one-component and two-component systems suffices. Our unified UNEP-v1 model exhibits superior performance across various physical properties compared to a widely used embedded-atom method potential, while maintaining remarkable efficiency. We demonstrate our approach’s effectiveness through reproducing experimentally observed chemical order and stable phases, and large-scale simulations of plasticity and primary radiation damage in MoTaVW alloys.