ScAClc: A Multi-Objective Adaptive Clustering Framework for Single-Cell Transcriptomics via Contrastive and Resolution-Aware Representation Learning.

Single-cell RNA sequencing (scRNA-seq) has revolutionized the field of cellular diversity research. Unsupervised clustering, a key technique in this exploration, allows for the identification of distinct cell types within a population. Graph-based deep clustering methods have shown promise in preserving the structural relationships between cells (nodes) within the data. However, these methods often neglect the inherent distribution of nodes in the graph, leading to incomplete representations of cell populations. Additionally, conventional graph convolutional networks (GCNs) can suffer from oversmoothing, a phenomenon where the network loses the ability to differentiate between samples with similar expression profiles. To address these limitations, we proposed scG-cluster, an innovative deep structural clustering method. This method incorporates two key innovations: (1) Dual-topology adjacency graph: scG-cluster integrates information about node distribution into the traditional adjacency graph used by GCNs. This enriches the graph representation by capturing the spatial relationships between cells in addition to their pairwise similarities. (2) Dual-topology adaptive graph convolutional network (TAGCN): The framework employs a TAGCN architecture with residual concatenation. This network utilizes an attention mechanism to dynamically weight features within the graph, focusing on the most informative aspects for clustering. Additionally, residual connections are implemented to combat oversmoothing, ensuring the network retains the ability to distinguish between subtle differences in cell expression profiles. Furthermore, scG-cluster iteratively refines the clustering centers, leading to enhanced stability and accuracy in the final cluster assignments. Extensive evaluations on six diverse scRNA-seq datasets demonstrate that scG-cluster consistently outperforms existing state-of-the-art methods in terms of both clustering accuracy and scalability. Ablation studies are also conducted to validate the significant contributions of both the residual connections and the attention mechanism to the overall performance of the model. The source code for scG-cluster is publicly available at https://github.com/xixi-wq/scG-cluster.

In recent years, single-cell RNA sequencing reveals diverse cell genetics at unprecedented resolutions. Such technological advances enable researchers to uncover the functionally distinct cell subtypes such as hematopoietic stem cell subpopulation identification. However, most of the related algorithms have been hindered by the high-dimensionality and sparse nature of single-cell RNA sequencing (RNA-seq) data. To address those problems, we propose a multiobjective evolutionary clustering based on adaptive non-negative matrix factorization (MCANMF) for multiobjective single-cell RNA-seq data clustering. First, adaptive non-negative matrix factorization is proposed to decompose data for feature extraction. After that, a multiobjective clustering algorithm based on learning vector quantization is proposed to analyze single-cell RNA-seq data. To validate the effectiveness of MCANMF, we benchmark MCANMF against 15 state-of-the-art methods including seven feature extraction methods, seven clustering methods, and the kernel-based similarity learning method on six published single-cell RNA sequencing datasets comprehensively. When compared with those 15 state-of-the-art methods, MCANMF performs better than the others on those single-cell RNA sequencing datasets according to multiple evaluation metrics. Moreover, the MCANMF component analysis, time complexity analysis, and parameter analysis are conducted to demonstrate various properties of our proposed algorithm.

Clustering is pivotal in deciphering cellular heterogeneity in single-cell RNA sequencing (scRNA-seq) data. However, it suffers from several challenges in handling the high dimensionality and complexity of scRNA-seq data. Especially when employing graph neural networks (GNNs) for cell clustering, the dependencies between cells expand exponentially with the number of layers. This results in high computational complexity, negatively impacting the model's training efficiency. To address these challenges, we propose a novel approach, called decoupled GNNs, based on multi-view contrastive learning (scDeGNN), for scRNA-seq data clustering. Firstly, this method constructs two adjacency matrices to generate distinct views, and trains them using decoupled GNNs to derive the initial cell feature representations. These representations are then refined through a multilayer perceptron and a contrastive learning layer, ensuring the consistency and discriminability of the learned features. Finally, the learned representations are fused and applied to the cell clustering task. Extensive experimental results on nine real scRNA-seq datasets from various organisms and tissues show that the proposed scDeGNN method significantly outperforms other state-of-the-art scRNA-seq data clustering algorithms across multiple evaluation metrics.

The rapid development of single-cell RNA sequencing (scRNA-seq) technology has provided unprecedented opportunities to explore cell heterogeneity and function. However, the high dimensionality, sparsity, and noise inherent in scRNA-seq data present significant challenges for traditional clustering methods. This review aims to summarize machine learningbased clustering techniques for scRNA-seq data, including Traditional Methods, Graph-based Methods, Ensemble Methods, Deep Learning Methods, and Other Methods, with a focus on discussing the advantages, limitations, and challenges of these approaches. We first discuss key preprocessing steps such as normalization, dropout imputation, and dimensionality reduction, which are essential for addressing data sparsity and improving clustering performance. Furthermore, the review introduces commonly used clustering performance evaluation metrics, including Adjusted Rand Index (ARI), Normalized Mutual Information (NMI), silhouette score, and marker gene validation. We also compare six distinct clustering methods across six datasets, evaluating the consistency in clustering accuracy with the selected methods. Our findings indicate that deep learning-based methods generally outperform other clustering methods in capturing complex relationships within the data, especially in high-dimensional and noisy datasets. However, challenges remain in areas such as computational efficiency, scalability for large-scale datasets, and handling batch effects. In this review, we systematically summarize the advantages and challenges of machine learning-based clustering algorithms. This work provides valuable insights and ideas for the development of new tools in the scRNA-seq clustering field and also helps address the numerous challenges faced in the downstream analysis of single- cell sequencing data.

OmniClust: A versatile clustering toolkit for single-cell and spatial transcriptomics data.

The development of technology in which robots autonomously learn the environment is remarkable. Although rewards are used for machine learning, robots operating in various environments require multiple evaluation metrics. This paper proposes a method to apply multiple evaluation metrics to policy gradient reinforce learning which is known as one of machine learning.

BackgroundRenal cell carcinoma (RCC) is a prevalent and aggressive form of kidney cancer, with variable response rates to checkpoint blockade immunotherapies. While high T cell infiltration often indicates a favorable...

SummaryExisting clustering methods for characterizing cell populations from single-cell RNA sequencing are constrained by several limitations stemming from the fact that clusters often cannot be homogeneous, particularly for transitioning populations. On the other hand, dominant cell populations within samples can be identified independently by their strong gene co-expression signatures using methods unrelated to partitioning. Here, we introduce a clustering method, CASCC (co-expression-assisted single-cell clustering), designed to improve biological accuracy using gene co-expression features identified using an unsupervised adaptive attractor algorithm. CASCC outperformed other methods as evidenced by multiple evaluation metrics, and our results suggest that CASCC can improve the analysis of single-cell transcriptomics, enabling potential new discoveries related to underlying biological mechanisms.Availability and implementationThe CASCC R package is publicly available at https://github.com/LingyiC/CASCC and https://zenodo.org/doi/10.5281/zenodo.10648327.

In this study, the uncertainty in runoff simulations using hydrological models was quantified based on the selection of five evaluation metrics and calibration data length. The calibration data length was considered to vary from 1 to 11 years, and runoff analysis was performed using a soil and water assessment tool (SWAT). SWAT parameter optimization was then performed using R-SWAT. The results show that the uncertainty was lower when using a calibration data length of five to seven years, with seven years achieving the lowest uncertainty. Runoff simulations using a calibration data length of more than seven years yielded higher uncertainty overall but lower uncertainty for extreme runoff simulations compared to parameters with less than five years of calibration data. Different uncertainty evaluation metrics show different levels of uncertainty, which means it is necessary to consider multiple evaluation metrics rather than relying on any one single metric. Among the evaluation metrics, the Nash–Sutcliffe model efficiency coefficient (NSE) and normalized root-mean-squared error (NRMSE) had large uncertainties at short calibration data lengths, whereas the Kling–Gupta efficiency (KGE) and Percent Bias (Pbias) had large uncertainties at long calibration data lengths.

Understanding cell differentiation and development dynamics is key for single-cell transcriptome analysis. Current cell differentiation trajectory inference algorithms face challenges such as high dimensionality, noise, and a need for users to possess certain biological information about the datasets to effectively utilize the algorithms. Here, we introduce Trajectory Inference with Cell-Cell Interaction (TICCI), a novel way to address these challenges by integrating intercellular communication information. In recognizing crucial intercellular communication during development, TICCI proposes Cell-Cell Interactions (CCI) at single-cell resolution. We posit that cells exhibiting higher gene expression similarity patterns are more likely to exchange information via biomolecular mediators. TICCI is initiated by constructing a cell-neighborhood matrix using edge weights composed of intercellular similarity and CCI information. Louvain partitioning identifies trajectory branches, attenuating noise, while single-cell entropy (scEntropy) is used to assess differentiation status. The Chu-Liu algorithm constructs a directed least-square model to identify trajectory branches, and an improved diffusion fitted time algorithm computes cell-fitted time in nonconnected topologies. TICCI validation on single-cell RNA sequencing (scRNA-seq) datasets confirms the accuracy of cell trajectories, aligning with genealogical branching and gene markers. Verification using extrinsic information labels demonstrates CCI information utility in enhancing accurate trajectory inference. A comparative analysis establishes TICCI proficiency in accurate temporal ordering. Source code and binaries freely available for download at https://github.com/mine41/TICCI, implemented in R (version 4.32) and Python (version 3.7.16) and supported on MS Windows. Authors ensure that the software is available for a full two years following publication.

To address the limitations of traditional computational fluid dynamics (CFD) simulations, such as high computational cost, long processing times, and limited scalability, this study identifies the inefficiencies of existing data‐driven prediction methods, which often lack spatial–temporal coordination mechanisms and fail to capture fine‐grained dynamic features of UAV airflow fields. We propose a novel deep learning model, VAN‐ConvLSTM, for rapid and accurate prediction of UAV downwash airflow. Unlike conventional ConvLSTM‐based frameworks, which struggle with modeling long‐range dependencies and detailed spatial variations, our model introduces a visual attention unit (VAN) to enhance spatiotemporal sensitivity. The model architecture combines a convolutional encoder for spatial feature extraction, a VAN module for attention‐guided temporal modeling, and a ConvLSTM decoder for sequence generation. This synergistic design improves both the accuracy and interpretability of airflow prediction. Experimental results show that the VAN‐ConvLSTM model achieves an SSIM score of 0.96, demonstrating high consistency with CFD simulations. Compared to baseline methods, our model reduces error while improving stability and spatial fidelity. Ablation studies further validate the individual contributions of VAN and ConvLSTM modules. The results, verified through three representative case studies, confirm that VAN‐ConvLSTM outperforms state‐of‐the‐art approaches across multiple evaluation metrics, while offering significantly enhanced computational efficiency. This demonstrates its strong potential as a reliable and scalable alternative to traditional CFD methods in rotor airflow prediction scenarios.

This study proposes a multi-stage intelligent diagnostic approach integrating Physics-Guided Normalization (LPGN), enhanced Transformer networks, and Gaussian Mixture Models (GMM) for thermal fault detection in turbine generator stators. The methodology sequentially performs the following steps: (1) enhances localized anomaly features in temperature data through LPGN, (2) efficiently extracts temporal patterns via the optimized Transformer architecture, and (3) achieves unsupervised fault classification using GMM. Experimental results demonstrate the proposed method's superiority over conventional ARIMA and LSTM models across multiple evaluation metrics, exhibiting a lower RMSE and a higher detection accuracy. Ablation studies further validate the individual contributions of each component to performance improvement. This solution provides an efficient and reliable framework for intelligent thermal monitoring in large rotating electrical machinery.

Explainable machine learning for predicting opioid-related aberrant behavior: A multimodal approach using clinical text and structured data.

Timely and accurate detection of agricultural disasters is crucial for ensuring food security and enhancing post-disaster response efficiency. This paper proposes a deployable UAV-based multimodal agricultural disaster detection framework that integrates multispectral and RGB imagery to simultaneously capture the spectral responses and spatial structural features of affected crop regions. To this end, we design an innovative stride–cross-attention mechanism, in which stride attention is utilized for efficient spatial feature extraction, while cross-attention facilitates semantic fusion between heterogeneous modalities. The experimental data were collected from representative wheat and maize fields in Inner Mongolia, using UAVs equipped with synchronized multispectral (red, green, blue, red edge, near-infrared) and high-resolution RGB sensors. Through a combination of image preprocessing, geometric correction, and various augmentation strategies (e.g., MixUp, CutMix, GridMask, RandAugment), the quality and diversity of the training samples were significantly enhanced. The model trained on the constructed dataset achieved an accuracy of 93.2%, an F1 score of 92.7%, a precision of 93.5%, and a recall of 92.4%, substantially outperforming mainstream models such as ResNet50, EfficientNet-B0, and ViT across multiple evaluation metrics. Ablation studies further validated the critical role of the stride attention and cross-attention modules in performance improvement. This study demonstrates that the integration of lightweight attention mechanisms with multimodal UAV remote sensing imagery enables efficient, accurate, and scalable agricultural disaster detection under complex field conditions.

Accurate and timely diagnosis of cardiovascular diseases, particularly myocardial infarction (MI), remains a critical clinical challenge. Existing electrocardiogram (ECG) analysis methods often rely solely on a single data modality, such as raw signals or waveform images, which limits their ability to capture the broader physiological context. To address this limitation, we propose GFM-MIP, a Graph-informed and FiLM-enhanced Multimodal Fusion framework for myocardial infarction prediction. GFM-MIP integrates 12-lead ECG time-series signals, ECG images, and laboratory test results through a unified architecture. Specifically, it employs a Graphormer encoder to model inter-lead dependencies in ECG signals and a Vision Transformer to extract morphological patterns from ECG images, both modulated by patient-specific laboratory features using Feature-wise Linear Modulation (FiLM). A Transformer-based fusion module captures cross-modal interactions, while a contrastive learning objective encourages alignment between signal and image modalities. Experimental results on a real-world clinical dataset and three public benchmarks demonstrate that GFM-MIP consistently outperforms state-of-the-art baselines across multiple evaluation metrics. Ablation studies further validate the contribution of each modality and architectural component. The proposed framework offers a clinically meaningful and scalable solution for robust, multimodal cardiovascular diagnosis.