Interdisciplinary Sciences: Computational Life Sciences最新文献

Alzheimer's disease (AD) is a progressive neurodegenerative disorder with increasing prevalence among the ageing population, necessitating early and accurate diagnosis for effective disease management. In this study, we present a novel hybrid deep learning framework, AlzhiNet, that integrates both 2D convolutional neural networks (2D-CNNs) and 3D convolutional neural networks (3D-CNNs), along with a custom loss function and volumetric data augmentation, to enhance feature extraction and improve classification performance in AD diagnosis. According to extensive experiments, AlzhiNet outperforms standalone 2D and 3D models, highlighting the importance of combining these complementary representations of data. The depth and quality of 3D volumes derived from the augmented 2D slices also significantly influence the model's performance. The results indicate that carefully selecting weighting factors in hybrid predictions is imperative for achieving optimal results. Our framework has been validated on the magnetic resonance imaging (MRI) from Kaggle and MIRIAD datasets, obtaining accuracies of 98.9% and 99.99%, respectively, with an AUC of 100%. Furthermore, AlzhiNet was studied under a variety of perturbation scenarios on the Alzheimer's Kaggle dataset, including Gaussian noise, brightness, contrast, salt and pepper noise, color jitter, and occlusion. The results obtained show that AlzhiNet is more robust to perturbations than ResNet-18, making it an excellent choice for real-world applications. This approach represents a promising advancement in the early diagnosis and treatment planning for AD.

Despite the current developments in mathematical modelling of biological process, some phenomena such as those encountered with the aspects of cell populations remain poorly understood. Fractional differential equations (FDEs) recently have received a significant amount of attention and demonstrated its rigor in representing real-world problems as opposed to traditional differential equations. In the present work, a systematic investigation using a mathematical approach dealing with the effects of ionizing radiation and using FDEs is proposed to illuminate some biological properties of the cell populations. For this purpose, the theoretical revelation of the cells population memory was treated within the context of FDEs, where the Mittag-Leffler function and Caputo derivatives are used to consider genetic potentials and memory traces. The model verification based on the parameter estimation algorithms is then accomplished by the implementation of two evolutionary hybrid optimization methods, namely the genetic algorithm-sequential quadratic programming (GA-SQP) and the particle swarm optimization-sequential quadratic programming (PSO-SQP). These algorithms have recently gained prominence as they present a practical approach to managing cell populations as well as their ability to effectively estimate the quality of the proposed solution by achieving the optimal solution. Insights and knowledge derived from the optimization of the objective function used in these two algorithms, whether through maximization or minimization, significantly contribute to the enhancement of evolutionary computation within the same cell population. The performance of these two algorithms is illustrated by determining the difference between the optimal results determined from GA-SQP and PSO-SQP algorithms. Both Control data and Bismuth Oxide Nanoparticles (BIONPS) survival experimental data are used. The reliability of the algorithms is elucidated based on the number of iterations, the computational time as well as the sum of squared error values. The linear quadratic method is used for treating the evolutionary computation of the cell population. By contrasting the theoretical findings with experimental results, it turns out that both PSO-SQP and GA-SQP optimization methods provide a correlation value close to experimental data and the estimated survival data. This emerging methodology reliably demonstrates the capability of the model to accurately fit the experimental data. Interestingly, a greater efficiency and effectiveness of the proposed PSO-SQP algorithm than the GA-SQP algorithm is observed suggesting hence the superiority of the PSO-SQP algorithm for determining the most realistic estimates of all the six model parameters studied herein.

Conventional approaches to drug discovery often require considerable time and effort. The promising solution is to repurpose existing drugs by identifying new therapeutic roles, thereby enhancing development efficiency. Drug repositioning based on computational methods is gaining widespread attention. However, most computational methods primarily rely on similarity-based data to extract features of associations, but lack the mining of topological structural features in the association network, while ignoring valuable original biological and chemical information. Therefore, this article develops a drug repositioning approach via meta-path integration of multi-source biological information (MPMB-DR). This approach combines meta-path and biomolecular similarity information to construct high-quality negative links within heterogeneous networks. It considers both the topological structure of the association network and the relationships among biomolecules. Based on the negative sample strategy, potential drug-disease associations are predicted by leveraging the synergy between meta-paths and multi-source biological data. Experimental results and case studies demonstrate that the MPMB-DR method has significant advantages in identifying associations between potential drugs and diseases.

Quorum sensing regulates cooperative behaviors in bacteria through the accumulation and detection of signaling molecules. This process plays a crucial role in various biological functions, including biofilm formation, antibiotic production, regulation of virulence factors, and immune modulation. Quorum sensing peptides (QSPs), primarily produced by Gram-positive bacteria, are key components of the quorum sensing mechanism, and their identification is crucial for understanding bacterial regulation. Despite the availability of several QSP prediction tools based on handcrafted features and machine learning techniques, there is still potential for improving their performance and interpretability. In this study, we present IQSPred-PLM, a novel model for predicting QSPs that integrates protein language models (PLMs) with a convolutional neural network (CNN). First, we utilize the pre-trained PLM ESM-2 to encode peptide sequences. Then, feature extraction is performed using a multi-scale residual CNN (MSRes-CNN), with dynamic feature integration through an adaptive weight modulation (AWM) module. Finally, a fully connected network is designed to conduct the classification of QSPs. Evaluated on the benchmark dataset, IQSPred-PLM demonstrated the outstanding predictive performance with accuracy (ACC), Matthews correlation coefficient (MCC), and the area under the receiver operating characteristic (ROC) curve (AUC) of 97.50%, 0.951, and 0.990, respectively. Furthermore, case studies and interpretability analyses confirmed the effectiveness of IQSPred-PLM for the QSP prediction task.

Clear cell renal cell carcinoma (ccRCC) is the most common form of renal cell carcinoma in adults, comprising approximately 80% of cases. The lethality of ccRCC rises significantly at stage III or beyond, emphasizing the need for early detection to enable timely therapeutic interventions. This study introduces a non-invasive and efficient classification method, Domain Adaptive Squeeze-and-Excitation Network (DASNet), for grading ccRCC through Computed Tomography (CT) images using advanced deep learning and machine learning techniques. The dataset is enhanced using MedAugment technology and balanced to improve generalization and classification performance. To mitigate overfitting, renal angiomyolipoma (AML) samples are incorporated, increasing data diversity and model robustness. EfficientNet and RegNet serve as foundational models, leveraging local feature extraction and Squeeze-and-Excitation (SE) attention mechanisms to enhance recognition accuracy across grades. Furthermore, Domain-Adversarial Neural Networks (DANNs) are employed to maintain consistency between source and target domains, bolstering the model's generalization ability. The proposed model achieves a classification accuracy of 97.50%, demonstrating efficacy in early ccRCC grade identification. These findings not only offer valuable clinical insights but also establish a foundation for broader application of deep learning in tumor detection.

De novo assembly and genome binning are fundamental steps for genome-resolved metagenomics analyses. However, the availability of limited computational resources and extensive processing time limit the broader application of these analyses. To address these challenges, the optimization of the parameters employed in these processes can improve the effective utilization of available metagenomics tools. Therefore, this study tested three sets of k-mers (default, reduced, and extended) for their efficiency in metagenome assembly and suitability in recovering metagenome-assembled genomes. The results demonstrate that the reduced set of k-mers outperforms the other two sets in computational efficiency and the quality of results. The assemblies from the default set are comparable with those from the reduced set; however, less complete and highly contaminated metagenome-assembled genomes are obtained at the expense of higher processing time. The extended set of k-mers yields less contiguous but computationally expensive assemblies. This set takes approximately 3-times more processing time than the reduced k-mers and recovers the lowest proportions of high and medium-quality metagenome-assembled genomes. Contrarily, the reduced set produces better assemblies, substantially improving the number and quality of the recovered metagenome-assembled genomes in significantly reduced processing time. Validation of the reduced k-mer set on previously published metagenome datasets further demonstrates its effectiveness not only for human metagenomes but also for the metagenomes of environmental origin. These findings underscore that the reduced k-mer set is optimal for efficient metagenome analyses of varying complexities and origins. This optimization of the k-mer set used in metagenome assemblers significantly reduces computational time while improving the quality of the assemblies and recovered metagenome-assembled genomes. This efficient solution will facilitate the widespread application of genome-resolved analyses, even in resource-limited settings, and help the recovery of better-quality metagenome-assembled genomes for downstream analyses.

Anti-inflammatory peptides (AIPs) have emerged as potential therapeutic candidates for managing various inflammatory disorders, but their computational identification remains challenging. We propose AIP-TranLAC, a novel deep learning framework that integrates Transformer-based embedding, bidirectional long short-term memory (Bi-LSTM), multi-head attention, and convolutional neural network (CNN) to classify AIPs accurately. Our model achieves superior performance on benchmark and independent test datasets, demonstrating significant improvements over existing methods. The hybrid architecture effectively captures local and global sequence patterns, while interpretability analyses reveal critical amino acid residues. With robust performance on imbalanced data and open-source availability, AIP-TranLAC provides a powerful tool for accelerating therapeutic peptide discovery and inflammation research. For reproducibility purposes, we have released the codebase, trained models, and all supporting data on GitHub ( https://github.com/Renjingyi123/AIP-TranLAC ).

The identification of protein homologs in large databases is critical for biological advancements. Traditional methods, such as protein sequence alignment, often miss remote homologs. To address this limitation, we present the Basic Embedding Search Tool (BEST), a fast and sensitive approach that employs protein language models to create sequence embeddings enriched with evolutionary and structural information. Besides, we introduce a segmented distillation pruning technique to accelerate sequence encoding and develop a multi-layer acceleration structure to achieve a 4290.86-fold speedup in swift access and retrieval of dense vectors. Extensive experiments on real datasets demonstrate that BEST increases sensitivity by over 20% compared to prior methods while maintaining precision and recall. It operates 23.41 times faster than traditional tools like PSI-BLAST and 3.92 times faster than Foldseek, while also detecting homologous sequences that conventional methods miss. BEST and its open-access web server ( http://pm2s.cpolar.top/best1/ ) are poised to significantly aid enzyme mining and advance biological research. The code is publicly available at https://github.com/SkyTai-W/ProteinMiningEvaluator .

Missense mutations are common in the coding genome and can alter protein functions. Distinguishing pathogenic from benign variants remains challenging despite computational advances. In the present work, we introduce EMMVEP, an ensemble-based approach designed for predicting the effects of protein missense mutations. EMMVEP leverages categorical boosting to integrate different types of features: one-hot encoding from protein sequence, physicochemical and environment properties extracted from AlphaFold database, and allele frequency information from gnomAD. When evaluated on a benchmark dataset with 112,832 clinical significance labels, our method achieved AUC and AUPR of 0.907 and 0.879, outperforming 20 general VEP methods. To aid in the identification of pathogenic mutations among the vast number of rare variants discovered through large-scale sequencing studies, we provide the pathogenicity probabilities of 216 million potential amino acid substitutions in 19,233 human protein-encoding genes. Our work demonstrates that EMMVEP can offer valuable independent insights for missense mutation interpretation in proteins, with significant applicability in both research and clinical contexts.

Understanding circuit-level imbalances in the cortex can yield mechanistic insights into Alzheimer's disease (AD), supporting both diagnosis and therapeutic development. We present a computational framework that integrates the causal interpretability of mechanistic modeling with the predictive power of simulation-based inference (SBI) to identify candidate neuroimaging biomarkers of cortical circuit dysfunction in AD. Using a spiking cortical circuit model with recurrent excitatory and inhibitory populations, we generated a comprehensive dataset of two million simulations and produced realistic electroencephalography (EEG) signals through biophysically grounded causal filtering of spiking activity. From these signals, we extracted EEG features serving as potential biomarkers of cortical dysregulation and trained SBI models optimized for accuracy and efficiency. Comparisons across feature sets revealed that multi-feature SBI models achieved higher inference accuracy than single-feature approaches in predicting various cortical parameters, suggesting that no single biomarker is sufficient to fully characterize the neural processes underlying the EEG signal. Applying the best-performing models to real EEG data from AD patients at varying stages uncovered distinct patterns of cortical dysfunction, including a progressive reduction in cortico-cortical connectivity, linked to the accelerated breakdown of synaptic connections widely reported in AD progression. A reduction in the efficacy of the excitatory time constant was also observed, likely reflecting a shift in the excitation/inhibition (E/I) balance toward inhibition in later stages of the disease. Our framework provides a scalable and interpretable bridge between local-scale mechanistic brain modeling and clinical neuroimaging, advancing the identification of physiologically meaningful biomarkers of cortical dysfunction in AD.