11563 research outputs found
Sort by
Biological AIE Molecules: Innovations in Synthetic Design and AI-Driven Discovery
No full textBiological aggregation -induced emission (AIE) molecules offer significant advantages over synthetic organic fluorophores, particularly in biocompatibility, environmental sustainability, and emission properties in biological systems. Derived from biomolecules such as peptides, proteins, and nucleic acids, biological AIE molecules hold great promise for applications in biosensing, bioimaging, and target drug delivery. This review explores the design principles, mechanistic insights, and functional properties of biological AIE molecules whiles highlighting the role of artificial intelligence (AI) in accelerating their discovery and optimization. AI-driven approaches, including machine learning and computational modeling, are transforming the identification and synthesis of AIE molecules by enabling precise structural modifications and enhanced fluorescence efficiency. These advancements are paving the way for the integration of AIE molecules in next-generation smart biomedical devices, personalized medicine and sustainable technological applications. Emerging trends, including hybrid biomaterials, Ai-guided molecular engineering, and advanced imaging techniques, are expanding the scope of biological AIE molecules in healthcare and environmental monitoring. The synergy between AI and biological AIE molecules is unlocking new frontiers in biomedical technology, enabling transformative advancements in material science and healthcare applications, and shaping the future of fluorescence- based diagnostics and therapeutics
A Tree-Structured Neural Network Model for Household Energy Breakdown
No full textResidential buildings constitute roughly one-fourth of the total energy use across the globe. Numerous studies have shown that providing an energy breakdown increases residents' awareness of energy use and can help save up to 15% energy. A significant amount of prior work has looked into source-separation techniques collectively called non-intrusive load monitoring (NILM), and most prior NILM research has leveraged high-frequency household aggregate data for energy breakdown. However, in practice most smart meters only sample hourly or once every 15 minutes, and existing NILM techniques show poor performance at such a low sampling rate. In this paper, we propose a TreeCNN model for energy breakdown on low frequency data. There are three key insights behind the design of our model: i) households consume energy with regular temporal patterns, which can be well captured by filters learned in CNNs; ii) tree structure isolates the pattern learning of each appliance that helps avoid magnitude variance problem, while preserves relationship among appliances; iii) tree structure enables the separation of known appliance from unknown ones, which de-noises the input time series for better appliance-level reconstruction. Our TreeCNN model outperformed seven existing baselines on a public benchmark dataset with lower estimation error and higher accuracy on detecting the active states of appliances
Comparing PAS domain coupled intrinsic dynamics in bHLH PAS domain transcription factor complexes
No full textThe basic Helix-Loop-Helix�Per-Arnt-Sim (bHLH-PAS) transcription factors (TFs) are regulators of several critical cellular functions such as circadian rhythm, hypoxia response and neuronal development. These proteins contain tandemly repeated PAS domains that mediate heterodimer formation. While PAS domains adopt a conserved fold, recent studies suggest that their interaction interfaces differ distinctly in different TF complexes. However, the implications of these differences on the intrinsic dynamics of PAS domains remain unclear. In this study, we performed a comparative analysis of PAS domain dynamics across multiple bHLH-PAS TF complexes using all-atom Elastic Network Models (ENMs) and molecular dynamics (MD) simulations. We decomposed the intrinsic dynamics of PAS domains into self-coupled (internal domain dynamics) and directly coupled (interaction partner-influenced dynamics) motions using a projection-based approach. Our results show that self-coupled motions are more conserved across PAS domains than structure or sequence alone, while directly coupled motions capture the context-specific influence of partner proteins. Furthermore, hierarchical clustering of the overall covariance-based similarity scores revealed distinct grouping of CLOCK:BMAL1-type and HIF:ARNT-type complexes, which were not captured by sequence or structural comparisons. Root mean square fluctuation profiles derived from both MD and ENM approaches showed strong correspondence, validating the utility of ENMs in capturing biologically relevant dynamics, even in cases where the structural complexes were modelled using AlphaFold3. PAS-B domains were generally found to be less flexible than PAS-A domains for all the complexes analysed. Regions with high directly coupled flexibility were generally localized regions with high interface propensity in class I PAS-B domains, suggesting a higher level of coupled dynamics between PAS-B domains. Our results highlight how PAS domain intrinsic dynamics are shaped by both their internal architecture and complex-specific interactions, offering new insights into the functional diversification of bHLH-PAS transcription factors.
Statement of Significance PAS domains are ubiquitous across all domains of life with diverse functions attributed to them. As part of bHLH-PAS transcription factors (TFs), they enable dimerization of Class I and Class II TFs. In this work, we investigated the effect of dimerization of PAS domains on their flexibility by using a method that allows us to isolate the intrinsic dynamics internal to a target domain and the intrinsic dynamics linked to the crosstalk between domains, from the all-atom elastic network model-based normal modes of the whole TF complex. Our findings reveal a context specific conservation of intrinsic dynamics based on the type of heterodimer complex. We also find a strong agreement between more-detailed MD simulations and the coarse-grained method used
Wormholes in finite cutoff JT gravity: a study of baby universes and (Krylov) complexity
No full textIn this paper, as an application of the `Complexity = Volume' proposal, we calculate the growth of the interior of a black hole at late times for finite cutoff JT gravity. Due to this integrable, irrelevant deformation, the spectral properties are modified non-trivially. The Einstein-Rosen Bridge (ERB) length saturates faster than pure JT gravity. We comment on the possible connection between Krylov Complexity and ERB length for deformed theory. Apart from this, we calculate the emission probability of baby universes for the deformed theory and make remarks on its implications for the ramp of the Spectral Form Factor. Finally, we compute the correction to the volume of the moduli space due to the non-perturbative change of the spectral curve because of the finite cutoff at the boundary
Small Molecule-Mediated Photothermal Therapy Induces Apoptosis in Cancer Cells
No full textCancer remains as one of the most life-threatening diseases in the whole world. Most of the therapeutic strategies to eradicate cancer are highly invasive, leading to severe injury and trauma to the patients. In recent times, phototherapy has emerged as one of the noninvasive therapeutic strategies for cancer treatment. However, development of novel small-molecule photothermal agents remains a major challenge. To address this, herein, a small molecule library having aromatic substituted-3-methoxy-pyrrole and 2-(3-cyano-4,5,5-trimethylfuran-2(5 H)-ylidene) malononitrile in a concise synthetic strategy is designed and synthesized. One of the library members (7H) self-assembles into spherical-like nanoparticles having <100 nm size in water and is found to exhibit remarkable increase in temperature under 740 nm near-infrared (NIR) light. Interestingly, compound 7H homes into the lysosomal compartments and the lipid droplets in the HCT-116 colon cancer cells within 3 h and induces photothermal effect followed by generation of reactive oxygen species while irradiating under 740 nm NIR light for 10 min. Moreover, 7H triggers programmed cell death (apoptosis) to induce remarkable HCT-116 cell killing. This small molecule-mediated photothermal effect shows potential to be an interesting tool for the next-generation noninvasive cancer phototherapy
Data-Driven Modeling of Li-Ion Battery Based on the Manufacturer Specifications and Laboratory Measurements
No full textAccurate modeling of Lithium-ion battery is essential in the development and testing of state estimation and lifetime prediction algorithms. The desired features of the model include flexibility, fast development, accuracy and reliability. There are many different ways to model a battery, depending on the level of abstraction desired, the data available and the target application environment. This paper shows how to extract equivalent circuit model parameters from manufacturer datasheets and laboratory measurement to build robust battery simulation models. A step-by-step methodology for data preparation is presented for both datasheet and measurement-based methods. The benefits and the disadvantages of both approaches are also discussed. A simple equivalent circuit model is firstly derived from manufacturer specification and its robustness is enhanced by collecting more extensive experimental data in the laboratory. Furthermore, an advanced model to better capture the battery dynamics is developed. The aging effects are added to this battery model, to reflect the internal parameters variation according to the health condition of the battery. To measure the accuracy of the developed models, the relative error is computed. An initial relative error of 2.8% of the model build with manufacturer specifications is reduced to 1.0% using laboratory measurements and finally to less than 0.4% by incorporating aging effects
Fault Tolerance of Oscillatory Neural Network: Device Oscillator-Based Small Network to Digital Oscillator-Based Large Networks
No full textOscillatory Neural Networks (ONN) are inevitable when it comes to solving combinatorial optimization problems. ONNs are also extremely energy efficient for AI workloads compared to conventional Deep Neural Networks (DNNs). Analysis of fault and failure tolerance of ONNs is crucial for understanding the reliability of the networks. This work illustrates the fault tolerance of the ONN in solving constraint optimization problems such as vertex coloring and digit recognition problems. For vertex coloring, a 4-node network across various configurations and different component failure levels has been analyzed using a device oscillator. The findings confirm that the network is highly robust to failures, demonstrating tolerance to variations in resistance of up to 40% and in capacitance of up to 60%. The analysis was then extended to bigger networks varying from 16-node network to 784-node network, using a digital oscillator for digit recognition of digits 0, 1, and 7. The results suggest that the tolerance shoots up rapidly as the network size increases, enhancing the stability of the ONN, making it highly robust. A saturation point exists beyond which the law of diminishing returns is observed. A tolerance of up to 99.9% in frequency fault and up to 59% in stuck-at fault is observed for extremely large networks of size 784 neurons
