Device quantization policy in variation-aware in-memory computing design

Device quantization of in-memory computing (IMC) that considers the non-negligible variation and finite dynamic range of practical memory technology is investigated, aiming for quantitatively co-optimizing system performance on accuracy, power, and area. Architecture- and algorithm-level solutions are taken into consideration. Weight-separate mapping, VGG-like algorithm, multiple cells per weight, and fine-tuning of the classifier layer are effective for suppressing inference accuracy loss due to variation and allow for the lowest possible weight precision to improve area and energy efficiency. Higher priority should be given to developing low-conductance and low-variability memory devices that are essential for energy and area-efficiency IMC whereas low bit precision (< 3b) and memory window (< 10) are less concerned.

Use of Adipose Stem Cells Against Hypertrophic Scarring or Keloid

Hypertrophic scars or keloid form as part of the wound healing reaction process, and its formation mechanism is complex and diverse, involving multi-stage synergistic action of multiple cells and factors. Adipose stem cells (ASCs) have become an emerging approach for the treatment of many diseases, including hypertrophic scarring or keloid, owing to their various advantages and potential. Herein, we analyzed the molecular mechanism of hypertrophic scar or keloid formation and explored the role and prospects of stem cell therapy, in the treatment of this condition.

Layered Feature Representation for Differentiable Architecture Search

Differentiable architecture search (DARTS) approach has made great progress in reducing the com- putational costs of neural architecture search. DARTS tries to discover an optimal architecture module called cell from a predefined super network. However, the obtained cell is then repeatedly and simply stacked to build a target network, failing to extract layered fea- tures hidden in different network depths. Therefore, this target network cannot meet the requirements of prac- tical applications. To address this problem, we propose an effective approach called Layered Feature Repre- sentation for Differentiable Architecture Search (LFR- DARTS). Specifically, we iteratively search for multiple cells with different architectures from shallow to deep layers of the super network. For each iteration, we optimize the architecture of a cell by gradient descent and prune out weak connections from this cell. After obtain- ing the optimal architecture of this cell, we deepen the super network by increasing the number of this cell, so as to create an adaptive network context to search for a deeper-adaptive cell in the next iteration. Thus, our LFR-DARTS can discover the architecture of each cell at a specific and adaptive network depth, which embeds the ability of layered feature representations into each cell to sufficiently extract layered features in different depths. Extensive experiments show that our algorithm achieves an advanced performance on the datasets of CIFAR10, fashionMNIST and ImageNet while at low search costs.

Hedgehog-inspired magnetic nanoparticles for effectively capturing and detecting exosomes

Exosomes, as one type of extracellular vesicle derived from multiple cells, have much potential as cancer biomarkers in clinical applications, but their enrichment and detection remain a huge challenge. Herein, inspired by the burr-like structure of the hedgehog, we present a new nanoparticle with a nanoneedle-assembled shell and a magnetic core for the effective capture and detection of exosomes. The unique nanoneedle structures endowed the magnetic nanoparticles with a large surface area for antibody modification so that the nanoparticles could serve as a platform for efficient exosome capture. In addition, the controllable movement of exosome-combined nanoparticles, which is due to the magnetic iron oxide cores, provides tremendous convenience for separating exosomes. The practical value of these nanoparticles in exosome analyses of serum from healthy patients and patients with liver cancer has also been demonstrated. Thus, we believe that bioinspired hierarchical nanoparticles are promising for biomedical and clinical applications.

Optical Fiber Tweezers for the Assembly of Living Photonic Probes

Optical fiber tweezers, as a versatile tool for optical trapping and manipulation, have attracted much attention in cell trapping, manipulation, and detection. Particularly, assembly of living cells using optical fiber tweezes has become a significant attention. Advanced achievements have been made on the assembly of fully biocompatible photonic probes with biological cells, enabling optical detection in biological environment in a highly compatible manner. Therefore, in this chapter, we discuss the use of optical fiber tweezers for assembly of living photonic probes. Living photonic probes can be assembled by the trapping and assembly of multiple cells using optical fiber tweezers. These photonic probes exhibit high biocompatibility and show great promise for the bio-applications in bio-microenvironments.

All-optical electrophysiology with improved genetically encoded voltage indicators reveals interneuron network dynamics in vivo

All-optical electrophysiology can be a powerful tool for studying neural dynamics in vivo, as it offers the ability to image and perturb membrane voltage in multiple cells simultaneously. The "Optopatch" constructs combine a red-shifted archaerhodopsin (Arch)-derived genetically encoded voltage indicator (GEVI) with a blue-shifted channelrhodopsin actuator (ChR). We used a video-based pooled screen to evolve Arch-derived GEVIs with improved signal-to-noise ratio (QuasAr6a) and kinetics (QuasAr6b). By combining optogenetic stimulation of individual cells with high-precision voltage imaging in neighboring cells, we mapped inhibitory and gap junction-mediated connections, in vivo. Optogenetic activation of a single NDNF-expressing neuron in visual cortex Layer 1 significantly suppressed the spike rate in some neighboring NDNF interneurons. Hippocampal PV cells showed near-synchronous spikes across multiple cells at a frequency significantly above what one would expect from independent spiking, suggesting that collective inhibitory spikes may play an important signaling role in vivo. By stimulating individual cells and recording from neighbors, we quantified gap junction coupling strengths. Together, these results demonstrate powerful new tools for all-optical microcircuit dissection in live mice.

Designing a phononic crystal with large-size defect to enhance elastic wave energy localization and harvesting

Phononic crystal (PnC) has been proved for its manipulation and amplification of elastic waves. Using this characteristic of PnC to assist energy harvesting has remarkable effect. Generally, defect occurs when unit cell in PnC is replaced by another cell with different geometric or material properties, the output electric power of piezoelectric energy harvesting (PEH) devices will be significantly enhanced. In this study, a cross hole-type PnC-assisted PEH device with a large-size defect is presented by replacing several adjacent multiple cells with other cells. It is found that multiple peak voltages can be created within BG and multimodal energy harvesting can be performed. Compared with the defect mode composed of a small-size defect, energy localization and amplification of the proposed PnC leads to substantially enhancement of harvesting power after tailoring geometric parameters of a PEH device. This work will be expected to design PnC-assisted PEH devices in a reasonable way.

Intracellular ATP Concentration and Implication for Cellular Evolution

Crystalline lens and striated muscle exist at opposite ends of the metabolic spectrum. Lens is a metabolically quiescent tissue, whereas striated muscle is a mechanically dynamic tissue with high-energy requirements, yet both tissues contain millimolar levels of ATP (>2.3 mM), far exceeding their underlying metabolic needs. We explored intracellular concentrations of ATP across multiple cells, tissues, species, and domains to provide context for interpreting lens/striated muscle data. Our database revealed that high intracellular ATP concentrations are ubiquitous across diverse life forms including species existing from the Precambrian Era, suggesting an ancient highly conserved role for ATP, independent of its widely accepted view as primarily “metabolic currency”. Our findings reinforce suggestions that the primordial function of ATP was non-metabolic in nature, serving instead to prevent protein aggregation.

Lipid droplets are beneficial for rabies virus replication by facilitating viral budding

Rabies is an old zoonotic disease caused by rabies virus (RABV), but the pathogenic mechanism of RABV is still not completely understood. Lipid droplets have been reported to play a role in pathogenesis of several viruses. However, its role on RABV infection remains unclear. Here, we initially found that RABV infection upregulated lipid droplet (LD) production in multiple cells and mouse brains. After the treatment of atorvastatin, a specific inhibitor of LD, RABV replication in N2a cells decreased. Then we found that RABV infection could upregulate N-myc downstream regulated gene-1 (NDRG1), which in turn enhance the expression of diacylglycerol acyltransferase 1/2 (DGAT1/2). DGAT1/2 could elevate cellular triglycerides synthesis and ultimately promote intracellular LD formation. Furthermore, we found that RABV-M and RABV-G, which were mainly involved in the viral budding process, could colocalize with LDs, indicating that RABV might utilize LDs as a carrier to facilitate viral budding and eventually increase virus production. Taken together, our study reveals that lipid droplets are beneficial for RABV replication and their biogenesis is regulated via NDRG1-DGAT1/2 pathway, which provides novel potential targets for developing anti-RABV drugs. IMPORTANCE Lipid droplets have been proven to play an important role in viral infections, but its role in RABV infection has not yet been elaborated. Here, we find that RABV infection upregulates the generation of LDs by enhancing the expression of N-myc downstream regulated gene-1 (NDRG1). Then NDRG1 elevated cellular triglycerides synthesis by increasing the activity of diacylglycerol acyltransferase 1/2 (DGAT1/2), which promotes the biogenesis of LDs. RABV-M and RABV-G, which are the major proteins involved in viral budding, could utilize LDs as a carrier and transport to cell membrane, resulting in enhanced virus budding. Our findings will extend the knowledge of lipid metabolism in RABV infection and help to explore potential therapeutic targets for RABV.