Direct training high-performance deep spiking neural networks: a review of theories and methods.

Spiking neural networks (SNNs) offer a promising energy-efficient alternative to artificial neural networks (ANNs), in virtue of their high biological plausibility, rich spatial-temporal dynamics, and event-driven computation. The direct training algorithms based on the surrogate gradient method provide sufficient flexibility to design novel SNN architectures and explore the spatial-temporal dynamics of SNNs. According to previous studies, the performance of models is highly dependent on their sizes. Recently, direct training deep SNNs have achieved great progress on both neuromorphic datasets and large-scale static datasets. Notably, transformer-based SNNs show comparable performance with their ANN counterparts. In this paper, we provide a new perspective to summarize the theories and methods for training deep SNNs with high performance in a systematic and comprehensive way, including theory fundamentals, spiking neuron models, advanced SNN models and residual architectures, software frameworks and neuromorphic hardware, applications, and future trends.

Auto-Spikformer: Spikformer architecture search.

Introduction: The integration of self-attention mechanisms into Spiking Neural Networks (SNNs) has garnered considerable interest in the realm of advanced deep learning, primarily due to their biological properties. Recent advancements in SNN architecture, such as Spikformer, have demonstrated promising outcomes. However, we observe that Spikformer may exhibit excessive energy consumption, potentially attributable to redundant channels and blocks. Methods: To mitigate this issue, we propose a one-shot Spiking Transformer Architecture Search method, namely Auto-Spikformer. Auto-Spikformer extends the search space to include both transformer architecture and SNN inner parameters. We train and search the supernet based on weight entanglement, evolutionary search, and the proposed Discrete Spiking Parameters Search (DSPS) methods. Benefiting from these methods, the performance of subnets with weights inherited from the supernet without even retraining is comparable to the original Spikformer. Moreover, we propose a new fitness function aiming to find a Pareto optimal combination balancing energy consumption and accuracy. Results and discussion: Our experimental results demonstrate the effectiveness of Auto-Spikformer, which outperforms the original Spikformer and most CNN or ViT models with even fewer parameters and lower energy consumption.

Correlating fluorescence microscopy, optical and magnetic tweezers to study single chiral biopolymers such as DNA.

Biopolymer topology is critical for determining interactions inside cell environments, exemplified by DNA where its response to mechanical perturbation is as important as biochemical properties to its cellular roles. The dynamic structures of chiral biopolymers exhibit complex dependence with extension and torsion, however the physical mechanisms underpinning the emergence of structural motifs upon physiological twisting and stretching are poorly understood due to technological limitations in correlating force, torque and spatial localization information. We present COMBI-Tweez (Combined Optical and Magnetic BIomolecule TWEEZers), a transformative tool that overcomes these challenges by integrating optical trapping, time-resolved electromagnetic tweezers, and fluorescence microscopy, demonstrated on single DNA molecules, that can controllably form and visualise higher order structural motifs including plectonemes. This technology combined with cutting-edge MD simulations provides quantitative insight into complex dynamic structures relevant to DNA cellular processes and can be adapted to study a range of filamentous biopolymers.

Feasibility and risk assessment of heavy metals from low-temperature magnetic pyrolysis of municipal solid waste on a pilot scale.

Low-temperature magnetic pyrolysis (LMP) of municipal solid waste (MSW) was conducted in a pilot scale continuous reactor to investigate the distribution and transformation of heavy metals (HMs) in biochar. Environmental safety was evaluated by the risk assessment code (RAC) and the modified potential ecological risk index (MRI). Statistical analyses of HMs revealed that the total concentrations of HMs in biochar was higher than that in MSW and the exchangeable fraction of Cd in biochar under 200 °C and 250 °C were at high risk levels. Temperature increment indicates an increase in regular steps not only migrated more HMs into biochar, but also broke the immobilization of HMs, so resulted in higher environmental risks. The lowest direct toxicity to the environment was obtained by LMP at 200 °C. In light of the residual fraction and the high concentration of HMs in biochar produced in this work, it should be mixed with other uncontaminated plant waste for further application in agriculture. The results of economic assessment reveal that the value of net present value (NPV) and the internal rate of return (IRR) can be positive if high quality bio-products are produced with low operating costs. Optimized design of operation, feedstock and the investment are the key factors to improve the economic feasibility of LMP.

High-Speed Single-Molecule Tracking of CXCL13 in the B-Follicle.

Soluble factors are an essential means of communication between cells and their environment. However, many molecules readily interact with extracellular matrix components, giving rise to multiple modes of diffusion. The molecular quantification of diffusion in situ is thus a challenging imaging frontier, requiring very high spatial and temporal resolution. Overcoming this methodological barrier is key to understanding the precise spatial patterning of the extracellular factors that regulate immune function. To address this, we have developed a high-speed light microscopy system capable of millisecond sampling in ex vivo tissue samples and submillisecond sampling in controlled in vitro samples to characterize molecular diffusion in a range of complex microenvironments. We demonstrate that this method outperforms competing tools for determining molecular mobility of fluorescence correlation spectroscopy (FCS) and fluorescence recovery after photobleaching (FRAP) for evaluation of diffusion. We then apply this approach to study the chemokine CXCL13, a key determinant of lymphoid tissue architecture, and B-cell-mediated immunity. Super-resolution single-molecule tracking of fluorescently labeled CCL19 and CXCL13 in collagen matrix was used to assess the heterogeneity of chemokine mobility behaviors, with results indicating an immobile fraction and a mobile fraction for both molecules, with distinct diffusion rates of 8.4 ± 0.2 and 6.2 ± 0.3 µm2s-1, respectively. To better understand mobility behaviors in situ, we analyzed CXCL13-AF647 diffusion in murine lymph node tissue sections and observed both an immobile fraction and a mobile fraction with an example diffusion coefficient of 6.6 ± 0.4 µm2s-1, suggesting that mobility within the follicle is also multimodal. In quantitatively studying mobility behaviors at the molecular level, we have obtained an increased understanding of CXCL13 bioavailability within the follicle. Our high-speed single-molecule tracking approach affords a novel perspective from which to understand the mobility of soluble factors relevant to the immune system.

Single-molecule techniques in biophysics: a review of the progress in methods and applications.

Single-molecule biophysics has transformed our understanding of biology, but also of the physics of life. More exotic than simple soft matter, biomatter lives far from thermal equilibrium, covering multiple lengths from the nanoscale of single molecules to up to several orders of magnitude higher in cells, tissues and organisms. Biomolecules are often characterized by underlying instability: multiple metastable free energy states exist, separated by levels of just a few multiples of the thermal energy scale k B T, where k B is the Boltzmann constant and T absolute temperature, implying complex inter-conversion kinetics in the relatively hot, wet environment of active biological matter. A key benefit of single-molecule biophysics techniques is their ability to probe heterogeneity of free energy states across a molecular population, too challenging in general for conventional ensemble average approaches. Parallel developments in experimental and computational techniques have catalysed the birth of multiplexed, correlative techniques to tackle previously intractable biological questions. Experimentally, progress has been driven by improvements in sensitivity and speed of detectors, and the stability and efficiency of light sources, probes and microfluidics. We discuss the motivation and requirements for these recent experiments, including the underpinning mathematics. These methods are broadly divided into tools which detect molecules and those which manipulate them. For the former we discuss the progress of super-resolution microscopy, transformative for addressing many longstanding questions in the life sciences, and for the latter we include progress in 'force spectroscopy' techniques that mechanically perturb molecules. We also consider in silico progress of single-molecule computational physics, and how simulation and experimentation may be drawn together to give a more complete understanding. Increasingly, combinatorial techniques are now used, including correlative atomic force microscopy and fluorescence imaging, to probe questions closer to native physiological behaviour. We identify the trade-offs, limitations and applications of these techniques, and discuss exciting new directions.

Force Spectroscopy in Studying Infection.

Biophysical force spectroscopy tools-for example, optical tweezers, magnetic tweezers, atomic force microscopy-have been used to study elastic, mechanical, conformational and dynamic properties of single biological specimens from single proteins to whole cells to reveal information not accessible by ensemble average methods such as X-ray crystallography, mass spectroscopy, gel electrophoresis and so on. Here, we review the application of these tools on a range of infection-related questions from antibody-inhibited protein processivity to virus-cell adhesion. In each case, we focus on how the instrumental design tailored to the biological system in question translates into the functionality suitable for that particular study. The unique insights that force spectroscopy has gained to complement knowledge learned through population averaging techniques in interrogating biomolecular details prove to be instrumental in therapeutic innovations such as those in structure-based drug design.

Probing DNA interactions with proteins using a single-molecule toolbox: inside the cell, in a test tube and in a computer.

DNA-interacting proteins have roles in multiple processes, many operating as molecular machines which undergo dynamic meta-stable transitions to bring about their biological function. To fully understand this molecular heterogeneity, DNA and the proteins that bind to it must ideally be interrogated at a single molecule level in their native in vivo environments, in a time-resolved manner, fast enough to sample the molecular transitions across the free-energy landscape. Progress has been made over the past decade in utilizing cutting-edge tools of the physical sciences to address challenging biological questions concerning the function and modes of action of several different proteins which bind to DNA. These physiologically relevant assays are technically challenging but can be complemented by powerful and often more tractable in vitro experiments which confer advantages of the chemical environment with enhanced detection signal-to-noise of molecular signatures and transition events. In the present paper, we discuss a range of techniques we have developed to monitor DNA-protein interactions in vivo, in vitro and in silico. These include bespoke single-molecule fluorescence microscopy techniques to elucidate the architecture and dynamics of the bacterial replisome and the structural maintenance of bacterial chromosomes, as well as new computational tools to extract single-molecule molecular signatures from live cells to monitor stoichiometry, spatial localization and mobility in living cells. We also discuss recent developments from our laboratory made in vitro, complementing these in vivo studies, which combine optical and magnetic tweezers to manipulate and image single molecules of DNA, with and without bound protein, in a new super-resolution fluorescence microscope.

Superresolution imaging of single DNA molecules using stochastic photoblinking of minor groove and intercalating dyes.

As proof-of-principle for generating superresolution structural information from DNA we applied a method of localization microscopy utilizing photoblinking comparing intercalating dye YOYO-1 against minor groove binding dye SYTO-13, using a bespoke multicolor single-molecule fluorescence microscope. We used a full-length ∼49 kbp λ DNA construct possessing oligo inserts at either terminus allowing conjugation of digoxigenin and biotin at opposite ends for tethering to a glass coverslip surface and paramagnetic microsphere respectively. We observed stochastic DNA-bound dye photoactivity consistent with dye photoblinking as opposed to binding/unbinding events, evidenced through both discrete simulations and continuum kinetics analysis. We analyzed dye photoblinking images of immobilized DNA molecules using superresolution reconstruction software from two existing packages, rainSTORM and QuickPALM, and compared the results against our own novel home-written software called ADEMS code. ADEMS code generated lateral localization precision values of 30-40 nm and 60-70 nm for YOYO-1 and SYTO-13 respectively at video-rate sampling, similar to rainSTORM, running more slowly than rainSTORM and QuickPALM algorithms but having a complementary capability over both in generating automated centroid distribution and cluster analyses. Our imaging system allows us to observe dynamic topological changes to single molecules of DNA in real-time, such as rapid molecular snapping events. This will facilitate visualization of fluorescently-labeled DNA molecules conjugated to a magnetic bead in future experiments involving newly developed magneto-optical tweezers combined with superresolution microscopy.