Biodata Mining最新文献

Automated ontology annotation of scientific literature plays a critical role in knowledge management, particularly in fields like biology and biomedicine, where accurate concept tagging can enhance information retrieval, semantic search, and knowledge integration. Traditional models for ontology annotation, such as Recurrent Neural Networks (RNNs) and Bidirectional Gated Recurrent Units (Bi-GRUs), have been effective but limited in handling complex biomedical terminologies and semantic nuances. This study explores the potential of large language models (LLMs), including MPT-7B, Phi, BiomedLM, and Meditron, for improving ontology annotation, specifically with Gene Ontology (GO) concepts. We fine-tuned these models on the CRAFT dataset, assessing their performance in terms of F1 score, semantic similarity, memory usage, and inference speed. Our results show that while Bi-GRU baselines remain competitive in raw accuracy, LLMs offer complementary strengths. LLMs exhibit qualitatively higher semantic consistency in some cases, particularly when handling complex or multi-word ontology terms. However, these observations are exploratory and not statistically verified across all model types. However, resource requirements for LLMs are notably high, raising considerations about computational efficiency. Techniques like parameter-efficient fine-tuning (PEFT) and advanced prompting were explored to address these challenges, demonstrating potential in reducing computational demands while maintaining performance. Our findings suggest that while LLMs offer advantages in annotation accuracy, practical deployment should balance these benefits with resource costs. This research highlights the need for further optimization and domain-specific training to make LLMs a feasible choice for real-world biomedical ontology annotation tasks.

Motivation: Protein-protein interactions (PPIs) are central to many biological processes, influencing cellular functions and offering potential therapeutic insights. The increasing availability of genomic and proteomic data has propelled computational approaches, particularly machine learning (ML), to the forefront of PPI prediction, addressing limitations of traditional experimental methods such as scalability and cost. This paper presents an in-depth review of modern ensemble learning models for PPI prediction, specifically focusing on XGBoost, Gradient Boosting, LightGBM, and Random Forest. Ensemble learning models are particularly noteworthy for their ability to surpass traditional approaches by leveraging the combined strengths of multiple base learners. Unlike previous reviews that often concentrate on theoretical comparisons or a limited selection of methods, this study offers a balanced analysis that includes both empirical findings and experimental evaluations. The review examines recent advancements in the field and assesses these models based on critical factors such as scalability, interpretability, accuracy, and efficiency, providing a structured framework to highlight nuanced differences among the techniques.

Scientific contribution: The article evaluates several ML techniques for PPI detection, with LightGBM and XGBoost emerging as the most effective methods due to their high accuracy and computational efficiency, particularly for large and complex datasets. Gradient Boosting and Random Forest also demonstrate strong performance, with Gradient Boosting excelling in capturing non-linear relationships and Random Forest offering robust interpretability. LightGBM stands out for its scalability and efficiency, while XGBoost benefits from built-in regularization to reduce overfitting.

Results: Experimental evaluations across three benchmark datasets-DIP, HPRD, and STRING-demonstrated that LightGBM consistently achieved the highest performance among all models, with average accuracy reaching up to 86%, and top sensitivity, specificity, and precision scores. XGBoost followed closely, showing strong generalization and robustness due to its regularization capabilities. Gradient Boosting provided competitive accuracy but lagged slightly in computational efficiency. Random Forest, while the most interpretable, showed comparatively lower sensitivity but maintained solid precision, highlighting its strength in minimizing false positives. Overall, LightGBM and XGBoost outperformed other methods in handling large, complex, and imbalanced PPI datasets. These results affirm the suitability of advanced ensemble learning techniques for enhancing PPI prediction and offer practical guidance on selecting models based on performance priorities and application constraints.

Background: Rickettsia typhi is the causative agent of epidemic murine typhus and Rocky Mountain spotted fever. The infection can affect multiple vital organs, including the heart, lungs, kidneys, and brain. Doxycycline is the recommended treatment but inflammation, mal-response, and drug resistance may arise. No natural product inhibitors have been reported against this bacterium.

Aim: The objective of this study was to establish a potential connection between autoimmune disorders triggered by R. typhi, identify therapeutic targets within its core proteome, and explore novel natural product inhibitors from Streptomyces spp. that could potentially inhibit it.

Methodology: Complete proteomes of four publicly available R. typhi strains were used for pan-proteomic analysis. The fni gene product (Isopentenyl pyrophosphate isomerase) was selected as the potential drug target. Molecular docking of 607 Streptomyces-derived metabolites was performed, with top hits validated using DiffDock and Vinardo scoring. Additionally, the Absorption, Distribution, Metabolism, Excretion, and Toxicity properties of the leading compounds were assessed via pkCSM, and formulation characteristics optimized using FormulationAI.

Results: Out of the 803 core proteins, associations between 14 proteins were mined for autoimmune diseases (including psoriasis, rheumatoid arthritis, optic atrophy, uveitis, even-plus syndrome, Sjogren syndrome, inflammatory bowel disease, allergic rhinitis, systemic lupus erythematosus, sclerosis, Stevens-Johnson syndrome, toxic epidermal necrolysis, colitis etc.). 17 core proteins were predicted as druggable. ZINC01482946 demonstrated the strongest inhibitory potential, as confirmed by DiffDock scoring, convolutional neural network-based ranking, and Vinardo scoring. It demonstrated a stable configuration and exhibited a favorable pharmacokinetic profile, with bioavailability enhanced through cyclodextrin complexation.

Conclusion: To the best of our knowledge, this is the first report identifying human autoimmune associations with R. typhi and natural product inhibitors targeting the pathogen. ZINC01482946 shows potential as an effective inhibitor of R. typhi, while SBE-β-CD appears to be a promising cyclodextrin for improving its solubility and bioavailability.

Clinical trial number: Not applicable.

Plant leaf diseases (PLDs) can continue to be a significant problem in the agricultural sector, leading to significant losses in production and jeopardizing food security. Early detection is essential, and recent achievements in the domain of deep learning (DL) have made automated high-accuracy solutions possible. The most popular and commonly used of these is the You Only Look Once (YOLO) family of object detection models, which have been proposed to detect plant diseases in real time. This review presents a new and in-depth synthesis of YOLO-based methods, including YOLOv1 to YOLOv10 and the domain-specific variants, including CTB-YOLO (coriander), BED-YOLO (YOLOv10n), and RAG-augmented YOLOv8 (coffee). This work compares to previous surveys in that (i) it presents a structured dataset catalog containing information on size, resolution, disease classes, and limitations (such as imbalance and annotation problems); (ii) it provides comparative benchmarking analysis of performance measures (accuracy, precision, recall, F1-score, mean Average Precision, and frames per second) across versions of YOLO to illustrate trade-offs between speed and accuracy; and (iii) it gives forward-looking discussion on how (ii) open challenges and (iii) future research directions, including lightweight YOLO models to run on mobile. This review presents a summative reference and a new contribution to the progress of the YOLO-based PLD detection approach to sustainable agriculture.