Interdisciplinary Sciences: Computational Life Sciences最新文献

As a common malignancy of the lower respiratory tract, non-small cell lung cancer (NSCLC) represents a major oncological challenge globally, characterized by high incidence and mortality rates. Recent research highlights the critical involvement of somatic mutations in the onset and development of NSCLC. Stratification of NSCLC patients based on somatic mutation data could facilitate the identification of patients likely to respond to personalized therapeutic strategies. However, stratification of NSCLC patients using somatic mutation data is challenging due to the sparseness of this data. In this study, based on sparse somatic mutation data from 4581 NSCLC patients from the Memorial Sloan Kettering Cancer Center (MSKCC) database, we systematically evaluate the metabolic pathway activity in NSCLC patients through the application of network propagation algorithm and computational biology algorithms. Based on these metabolic pathways associated with prognosis, as recognized through univariate Cox regression analysis, NSCLC patients are stratified using the deep clustering algorithm to explore the optimal classification strategy, thereby establishing biologically meaningful metabolic subtypes of NSCLC patients. The precise NSCLC metabolic subtypes obtained from the network propagation algorithm and deep clustering algorithm are systematically evaluated and validated for survival benefits of immunotherapy. Our research marks progress towards developing a universal approach for classifying NSCLC patients based solely on somatic mutation profiles, employing deep clustering algorithm. The implementation of our research will help to deepen the analysis of NSCLC patients' metabolic subtypes from the perspective of tumor microenvironment, providing a strong basis for the formulation of more precise personalized treatment plans.

Immune checkpoint inhibitors (ICIs) have demonstrated significant clinical benefits in cancer treatment, but only a minority of patients exhibit favorable response, highlighting the importance of determining patients who will benefit from immunotherapy. Currently, patient datasets regarding immunotherapy response are scarce, while ample experiments can be performed on syngeneic mouse tumor models to generate valuable data. Therefore, how to effectively utilize mouse data to identify predictors of immunotherapy response and subsequently transfer relevant knowledge to predict human response to ICIs is a question worth studying. In this study, we propose a novel methodology to address this issue. Firstly, we identify gene modules associated with immunotherapy response from mouse tumor profiles based on cancer gene panels. Subsequently, these identified modules are employed to build prediction models for immunotherapy response based on mouse data. Furthermore, we transfer these models to predict ICIs responses of human cancer patients. Experimental results demonstrate that the gene modules identified from mouse data are reliable predictors of immunotherapy response. The mouse-based models built on these modules could be transferred to humans, effectively predicting drug responses and survival outcomes for cancer patients. Compared to conventional cancer biomarkers and existing prediction models based on mouse data, our method exhibits superior performance. These findings provide a valuable reference for further in-depth research on immunotherapy response prediction model based on mouse tumor profiles, with the potential for transfer applications in human cancer therapy.

Few-shot Biomedical Named Entity Recognition (BioNER) presents significant challenges due to limited training data and the presence of nested and discontinuous entities. To tackle these issues, a novel approach GKP-BioNER, Generation-based Few-Shot BioNER via Local Knowledge Index and Dual Prompts, is proposed. It redefines BioNER as a generation task by integrating hard and soft prompts. Specifically, GKP-BioNER constructs a localized knowledge index using a Wikipedia dump, facilitating the retrieval of semantically relevant texts to the original sentence. These texts are then reordered to prioritize the most semantically relevant content to the input data, serving as hard prompts. This helps the model to address challenges demanding domain-specific insights. Simultaneously, GKP-BioNER preserves the integrity of the pre-trained models while introducing learnable parameters as soft prompts to guide the self-attention layer, allowing the model to adapt to the context. Moreover, a soft prompt mechanism is designed to support knowledge transfer across domains. Extensive experiments on five datasets demonstrate that GKP-BioNER significantly outperforms eight state-of-the-art methods. It shows robust performance in low-resource and complex scenarios across various domains, highlighting its strength in knowledge transfer and broad applicability.

In recent years, computational methods such as machine learning and deep learning have been increasingly used to solve various bioinformatics problems related to protein sequence data, such as predicting protein interaction, protein function, subcellular location, and so on. The first crucial step in applying these methods is how to represent a protein sequence as an input feature vector, as the feature vector quality significantly impacts the performance of those methods. A range of protein sequence descriptors has been proposed to enhance the quality of protein sequence representation. Existing descriptors extract information that can be obtained from sequences, such as composition, distribution, spatial correlation between amino acids, and so on. However, improvements can still be made in spatial correlation to capture better sequence similarity, which is valuable for Protein-Protein Interaction (PPI) prediction tasks. In this study, our aim is to develop new descriptors based on six well-known sequence descriptors to improve the ability to represent protein sequences. We evaluate the performance of the new descriptors on various PPI datasets. The results demonstrate that the proposed descriptors outperform their original versions in terms of PPI prediction performance. This work also introduces ProtSeqDesc (protein sequence descriptors), a flexible Python package that includes 51 types of feature vectors, covering all proposed descriptors. The software package is aimed at meeting the demand for the application of computational methods in bioinformatics.

Predicting drug-target binding affinity (DTA) is an important step in the complex process of drug discovery or drug repositioning. A large number of computational methods proposed for the task of DTA prediction utilize single features of proteins to measure drug-protein or protein-protein interactions, ignoring multi-feature fusion between protein-related features (e.g., solvent accessibility, protein pockets, secondary structures, and distance maps, etc.). To address the aforementioned constraints, we propose a new network topology and multi-feature fusion based approach for DTA prediction (NTMFF-DTA), which deeply mines protein multiple types of data and propagates drug information across domains. Data in drug-target interactions are often sparse, and multi-feature fusion can enrich data information by integrating multiple features, thus overcoming the data sparsity problem to some extent. The proposed approach offers two main contributions: (1) constructing a relationship-aware GAT that selectively focuses on the connections between nodes and edges in the molecular graph to capture the more central roles of nodes and edges in DTA prediction and (2) constructing an information propagation channel between different feature domains of drug proteins to achieve the sharing of the importance weight of drug atoms and edges, and combining with a multi-head self-attention mechanism to capture residue-enhancing features. The NTMFF-DTA model was comparatively tested against several leading baseline technologies on commonly used datasets. Experimental show that NTMFF-DTA can effectively and accurately predict DTA and outperform existing comparative models.

The advancement of deep learning has driven extensive research validating the effectiveness of U-Net-style symmetric encoder-decoder architectures based on Transformers for medical image segmentation. However, the inherent design requiring attention mechanisms to compute token affinities across all spatial locations leads to prohibitive computational complexity and substantial memory demands. Recent efforts have attempted to address these limitations through sparse attention mechanisms. However, existing approaches employing artificial, content-agnostic sparse attention patterns demonstrate limited capability in modeling long-range dependencies effectively. We propose MFFBi-Unet, a novel architecture incorporating dynamic sparse attention through bi-level routing, enabling context-aware computation allocation with enhanced adaptability. The encoder-decoder module integrates BiFormer to optimize semantic feature extraction and facilitate high-fidelity feature map reconstruction. A novel Multi-scale Feature Fusion (MFF) module in skip connections synergistically combines multi-level contextual information with processed multi-scale features. Extensive evaluations on multiple public medical benchmarks demonstrate that our method consistently exhibits significant advantages. Notably, our method achieves statistically significant improvements, outperforming state-of-the-art approaches like MISSFormer by 2.02% and 1.28% Dice scores on respective benchmarks.

Accurate prediction of drug-drug interaction (DDI) is essential to improve clinical efficacy, avoid adverse effects of drug combination therapy, and enhance drug safety. Recently researchers have developed several computer-aided methods for DDI prediction. However, these methods lack the substructural features that are critical to drug interactions and are not effective in generalizing across domains and different distribution data. In this work, we present SAGAN, a domain adaptive interpretable substructure-aware graph attention network for DDI prediction. Based on attention mechanism and unsupervised clustering algorithm, we propose a new substructure segmentation method, which segments the drug molecule into multiple substructures, learns the mechanism of drug interaction from the perspective of interaction, and identifies important interaction regions between drugs. To enhance the generalization ability of the model, we improve and apply a conditional domain adversarial network to achieve cross-domain generalization by alternately optimizing the cross-entropy loss on the source domain and the adversarial loss of the domain discriminator. We evaluate and compare SAGAN with the state-of-the-art DDI prediction model on four real-world datasets for both in-domain and cross-domain scenarios, and show that SAGAN achieves the best overall performance. Moreover, the visualization results of the model show that SAGAN has achieved pharmacologically significant substructure extraction, which can help drug developers screen for some undiscovered local interaction sites, and provide important information for further drug structure optimization. The codes and datasets are available online at https://github.com/wyx2012/SAGAN .

Accurate tumor grading and regional identification of cervical tumors are important for diagnosis and prognosis. Traditional manual microscopy methods suffer from time-consuming, labor-intensive, and subjective bias problems, so tumor segmentation methods based on deep learning are gradually becoming a hotspot in current research. Cervical tumors have diverse morphologies, which leads to low similarity between the mask edge and ground-truth edge of existing semantic segmentation models. Moreover, the texture and geometric arrangement features of normal tissues and tumors are highly similar, which causes poor pixel connectivity in the mask of the segmentation model. To this end, we propose an end-to-end semantic consistency network with the edge learner and the connectivity enhancer, i.e., ERNet. First, the edge learner consists of a stacked shallow convolutional neural network, so it can effectively enhance the ability of ERNet to learn and represent polymorphic tumor edges. Second, the connectivity enhancer learns detailed information and contextual information of tumor images, so it can enhance the pixel connectivity of the masks. Finally, edge features and pixel-level features are adaptively coupled, and the segmentation results are additionally optimized by the tumor classification task as a whole. The results show that, compared with those of other state-of-the-art segmentation models, the structural similarity and the mean intersection over union of ERNet are improved to 88.17% and 83.22%, respectively, which reflects the excellent edge similarity and pixel connectivity of the proposed model. Finally, we conduct a generalization experiment on laryngeal tumor images. Therefore, the ERNet network has good clinical popularization and practical value.

Emerging research continues to reveal the fundamental contributions of microbial communities to maintaining human physiological balance and advancing drug discovery. However, established wet-lab investigation techniques require significant time and resources. Contemporary research efforts have predominantly concentrated on establishing robust computational architectures to predict microbe-drug associations. Our research establishes a neural network architecture that synthesizes heterogeneous biological relationships with attentional factorization machines (HAFMMDA) to predict undiscovered microbe-drug linkages. The initial step involves assembling a heterogeneous network architecture integrating three key components: microbe similarity networks, drug similarity networks, and established microbe-drug interaction networks. HAFMMDA utilizes HIN2vec to extract feature representations of microbe-drug pairs. Finally, it combines second-order feature interactions and attention mechanism to perform comprehensive prediction. Five-fold cross-validation results confirmed excellent predictive performance with an AUC score of 0.9805, demonstrating statistically significant improvements over five contemporary baseline approaches. These findings corroborate HAFMMDA's effectiveness in uncovering verified drug-microorganism associations while simultaneously predicting innovative therapeutic-microbe relationships.