Year: 2025
Type: article
Abstract: 蛋白质间相互作用(protein-protein interaction, PPI)是维系细胞正常生理功能的核心要素，明确参与代谢调控、信号转导、免疫应答等诸多关键生物学进程，与人类健康紧密相关。深入探究PPI对于阐释疾病发生发展机制、挖掘潜在药物靶点意义重大。传统PPI预测方法多依赖实验手段，存在通量低、假阳性高以及难以处理复杂互作模式等问题。随着人工智能技术蓬勃发展，深度学习凭借强大特征学习与模态融合能力，逐渐成为PPI预测新锐手段。鉴于此，本研究创新性提出一种多模态融合的深度神经网络模型，有机整合蛋白质序列与结构特征，充分挖掘二者关联模式，实现对蛋白质间潜在互作关系精准预测。经多个权威基准数据集验证，该模型在准确率、召回率、AUC值等核心评估指标上全面超越现有方法。这一成果有望为蛋白质互作网络解析提供高效工具，为细胞信号传导规律探索、疾病分子机制研究注入新动力，进而为精准医学发展与药物研发进程加速助力。Protein-protein interaction (PPI) is the core elements that maintain normal cellular physiological functions, clearly participating in many critical biological processes such as metabolic regulation, signal transduction, and immune response, which are closely related to human health. In-depth investigation of PPI is crucial for elucidating disease mechanisms and uncovering potential drug targets. Traditional PPI prediction methods often rely on experimental approaches, facing issues like low throughput, high false positives, and difficulty in handling complex interaction patterns. With the rapid development of artificial intelligence technology, deep learning has emerged as a new and powerful tool for PPI prediction, thanks to its strong feature learning and modal fusion capabilities. In light of this, our study innovatively proposes a multi-modal fusion deep neural network model that organically integrates protein sequence and structural features, fully exploring their associated patterns to achieve precise prediction of potential protein interactions. Validation on multiple authoritative benchmark datasets shows that this model outperforms existing methods comprehensively in core evaluation metrics such as accuracy, recall rate, and AUC value. This achievement is expected to provide an efficient tool for protein interaction network analysis, inject new momentum into the exploration of cellular signaling rules and disease molecular mechanisms, and accelerate the development of precision medicine and drug research.
Source: Hans Journal of Biomedicine
Author: 跃花 白
Language: English
Subfield: Molecular Biology
Domain: Life Sciences
Open Access status: hybrid
OpenAlex ID: W4410945666