提示
dgraphppi：融合大模型语义和接触图信息的ppi预测
首发时间：2025-02-24
摘要：蛋白质-蛋白质相互作用（ppi）在信号传导、代谢调节和环境响应等多种细胞过程中发挥着至关重要的作用。传统的ppi鉴定方法，如酵母双杂交和质谱分析，通常耗时且成本高昂。近年来，基于机器学习的计算方法在ppi预测方面受到了广泛的关注。尽管基于机器学习的计算方法提高了预测效率，但在如何进行高效地特征学习方面仍面临挑战，使得ppi预测性能遇到瓶颈。为了应对此挑战，本文提出了一种新的结构感知方法dgraph-ppi。该方法将alphafold和蛋白质预训练模型（esm2）整合到深度图卷积网络框架中，实现了蛋白质特征的高效深度抽取。本文提出的dgraph-ppi模型能充分利用蛋白质的序列信息、结构信息和深度学习的特征提取能力，显著提升了ppi预测的准确性。在酵母核心数据集上的实验结果显示，本文方法达到了98.51%的准确率。
关键词：
for information in english, please click here
dgraphppi: ppi prediction by integrating large model semantics and contact map information
abstract：protein-protein interaction (ppi) plays a crucial role in various cellular processes, including signal transduction, metabolic regulation, and environmental responses. traditional methods for ppi identification, such as yeast two-hybrid and mass spectrometry, are often time-consuming and costly. recently, machine learning-based computational methods have gained widespread attention in ppi prediction. while these methods have improved prediction efficiency, challenges in effectively learning features remain, leading to bottlenecks in ppi prediction performance. to address this issue, we propose a novel structure-aware method, dgraph-ppi. this deep convolutional network effectively aggregates neighborhood information and alleviates the common issue of over-smoothing in deep networks. the dgraph-ppi model utilizes protein sequence data, structural information, and the feature extraction power of deep learning, significantly enhancing ppi prediction accuracy. experimental results on the yeast core dataset demonstrate that our method achieves an accuracy of 98.51%. furthermore, the dgraph-ppi method, which combines esm2 with deep graph neural networks, provides a novel research approach for bioinformatics-related studies.
keywords：
基金：
论文图表：
引用
导出参考文献
no.****
动态公开评议
共计0人参与
勘误表
dgraphppi：融合大模型语义和接触图信息的ppi预测
评论
全部评论