Yunyin Li
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Yunyin Li has a total of 24 co-authors. Here are the top co-authors by number of shared publications:
Shudong Wang
China University of Petroleum (East China)
6 shared publicationsShanchen Pang
China University of Petroleum (East China)
5 shared publicationsChuanru Ren
4 shared publicationsYingye Liu
4 shared publicationsYuanyuan Zhang
4 shared publications- Tiyao Liu
College of Computer Science and Technology, Qingdao Institute of Software, China University of Petroleum, Qingdao 266580, China
3 shared publications
Latest Publications
- 23 September 2026
- Published by American Chemical Society (ACS) in ACS Synthetic Biology
- Vol. 15 (10) , 4276-4290
MASP-CMI: A Sequence-Derived Multi-View Attribute and Similarity-Prior Evidence Learning Framework for circRNA−miRNA Interaction Prediction
CircRNA−miRNA interactions (CMIs) regulate downstream gene expression through the miRNA sponge mechanism and contribute to the progression of various diseases. Although computational methods have advanced the prediction of potential CMIs, existing approaches still face challenges such as sparse association networks, limited single-sequence representations, and insufficient use of intrinsic RNA sequence evidence. To address these issues, we propose MASP-CMI, a sequence-derived multi-view attribute and similarity-prior evidence learning framework for CMI prediction. Instead of relying on known association networks for node information propagation, MASP-CMI jointly learns multi-view attribute evidence and sequence similarity priors from RNA sequences. The heterogeneous attribute-view evidence learning module adaptively integrates FCGR, CTD, and Doc2Vec features, while the similarity-prior evidence aggregation module enhances RNA representations within homogeneous RNA spaces. Cross-branch alignment and pair-level interaction fusion are further used to predict potential CMIs. Experiments on three benchmark datasets demonstrate that MASP-CMI achieves stable predictive performance, strong robustness, and favorable cold-start generalization. Module ablation experiments and case studies further confirm its effectiveness for potential CMI discovery and experimental prioritization.
- 23 September 2026
- Published by American Chemical Society (ACS) in ACS Synthetic Biology
- Vol. 15 (10) , 4276-4290
MASP-CMI: A Sequence-Derived Multi-View Attribute and Similarity-Prior Evidence Learning Framework for circRNA−miRNA Interaction Prediction
CircRNA−miRNA interactions (CMIs) regulate downstream gene expression through the miRNA sponge mechanism and contribute to the progression of various diseases. Although computational methods have advanced the prediction of potential CMIs, existing approaches still face challenges such as sparse association networks, limited single-sequence representations, and insufficient use of intrinsic RNA sequence evidence. To address these issues, we propose MASP-CMI, a sequence-derived multi-view attribute and similarity-prior evidence learning framework for CMI prediction. Instead of relying on known association networks for node information propagation, MASP-CMI jointly learns multi-view attribute evidence and sequence similarity priors from RNA sequences. The heterogeneous attribute-view evidence learning module adaptively integrates FCGR, CTD, and Doc2Vec features, while the similarity-prior evidence aggregation module enhances RNA representations within homogeneous RNA spaces. Cross-branch alignment and pair-level interaction fusion are further used to predict potential CMIs. Experiments on three benchmark datasets demonstrate that MASP-CMI achieves stable predictive performance, strong robustness, and favorable cold-start generalization. Module ablation experiments and case studies further confirm its effectiveness for potential CMI discovery and experimental prioritization.
- 24 July 2026
- Published by Elsevier in Neurocomputing
- Vol. 701
RGST: A relation-guided semantic transformer for dual-view drug–disease association prediction
- 24 July 2026
- Published by Elsevier in Neurocomputing
- Vol. 701
RGST: A relation-guided semantic transformer for dual-view drug–disease association prediction
- 28 July 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Consumer Electronics
- Vol. 71 (3) , 7771-7783
ALRTGCL: Adaptive Low-Rank Transformer and Graph Contrastive Learning for MiRNA–Drug Associations Prediction in Consumer Healthcare
Within the domain of consumer health electronics, precise prediction of miRNA–drug associations plays a vital role in enhancing therapeutic approaches and accelerating the drug discovery process. Traditional approaches often suffer from high computational costs and limited capability to capture intricate patterns in complex and heterogeneous datasets. To address these challenges, we present ALRTGCL, a novel framework designed to significantly enhance miRNA–drug association prediction by combining an Adaptive Low-Rank Transformer with Graph Contrastive Learning. The core innovation of ALRTGCL is its adaptive low-rank attention mechanism, which dynamically adjusts attention scores based on data characteristics, thereby reducing model complexity and computational overhead without compromising predictive accuracy. In addition, we incorporate graph contrastive learning to capture diverse graph-based perspectives of the data. This methodology refines the representation of miRNA–drug associations by exploiting subtle differences across multiple data views, thereby enhancing the model’s ability to capture complex relationships and improve its overall performance. We evaluate ALRTGCL through extensive experiments on multiple benchmark datasets, demonstrating its superior predictive accuracy and scalability. The results position ALRTGCL as a highly effective tool for personalized medicine and health monitoring applications, with its computational efficiency making it ideal for integration into medical electronic devices that promote healthy patient behaviors.
- 28 July 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Consumer Electronics
- Vol. 71 (3) , 7771-7783
ALRTGCL: Adaptive Low-Rank Transformer and Graph Contrastive Learning for MiRNA–Drug Associations Prediction in Consumer Healthcare
Within the domain of consumer health electronics, precise prediction of miRNA–drug associations plays a vital role in enhancing therapeutic approaches and accelerating the drug discovery process. Traditional approaches often suffer from high computational costs and limited capability to capture intricate patterns in complex and heterogeneous datasets. To address these challenges, we present ALRTGCL, a novel framework designed to significantly enhance miRNA–drug association prediction by combining an Adaptive Low-Rank Transformer with Graph Contrastive Learning. The core innovation of ALRTGCL is its adaptive low-rank attention mechanism, which dynamically adjusts attention scores based on data characteristics, thereby reducing model complexity and computational overhead without compromising predictive accuracy. In addition, we incorporate graph contrastive learning to capture diverse graph-based perspectives of the data. This methodology refines the representation of miRNA–drug associations by exploiting subtle differences across multiple data views, thereby enhancing the model’s ability to capture complex relationships and improve its overall performance. We evaluate ALRTGCL through extensive experiments on multiple benchmark datasets, demonstrating its superior predictive accuracy and scalability. The results position ALRTGCL as a highly effective tool for personalized medicine and health monitoring applications, with its computational efficiency making it ideal for integration into medical electronic devices that promote healthy patient behaviors.
- 17 June 2025
- Published by American Chemical Society (ACS) in Journal of Chemical Information and Modeling
- Vol. 65 (13) , 7302-7314
Higher-Order Weighted Perturbation-Based Multilevel Information Fusion Model for Predicting CircRNA–Disease Associations
CircRNAs are closely associated with the initiation and progression of various diseases, and investigating their potential associations is crucial for understanding disease mechanisms. Although existing computational methods have made notable progress, they often overlook the latent information contained in higher-order associations between circRNAs and diseases. Moreover, these methods often rely on a single linear or nonlinear feature learning approach, which fails to comprehensively capture multilevel features. To address this, a high-order weighted perturbation-based multilevel information fusion model (HWP-MIFM) is proposed. The model dynamically adjusts the weights of different orders using a higher-order weighted perturbation method to extract higher-order association information. A dual-stage matrix factorization module is applied to construct a multilayer structure and extract linear features. Additionally, a dual-path feature learning module is utilized to dig complex nonlinear relationships within similarity networks, ensuring the comprehensive capture of multilevel features. Experimental results demonstrate that, in 5-fold cross-validation on four data sets, HWP-MIFM outperforms seven state-of-the-art prediction methods in terms of overall performance. Ablation studies and case analyses further confirm the accuracy and practical value of the model.
- 17 June 2025
- Published by American Chemical Society (ACS) in Journal of Chemical Information and Modeling
- Vol. 65 (13) , 7302-7314
Higher-Order Weighted Perturbation-Based Multilevel Information Fusion Model for Predicting CircRNA–Disease Associations
CircRNAs are closely associated with the initiation and progression of various diseases, and investigating their potential associations is crucial for understanding disease mechanisms. Although existing computational methods have made notable progress, they often overlook the latent information contained in higher-order associations between circRNAs and diseases. Moreover, these methods often rely on a single linear or nonlinear feature learning approach, which fails to comprehensively capture multilevel features. To address this, a high-order weighted perturbation-based multilevel information fusion model (HWP-MIFM) is proposed. The model dynamically adjusts the weights of different orders using a higher-order weighted perturbation method to extract higher-order association information. A dual-stage matrix factorization module is applied to construct a multilayer structure and extract linear features. Additionally, a dual-path feature learning module is utilized to dig complex nonlinear relationships within similarity networks, ensuring the comprehensive capture of multilevel features. Experimental results demonstrate that, in 5-fold cross-validation on four data sets, HWP-MIFM outperforms seven state-of-the-art prediction methods in terms of overall performance. Ablation studies and case analyses further confirm the accuracy and practical value of the model.
- 21 February 2025
- Published by American Chemical Society (ACS) in Journal of Chemical Information and Modeling
- Vol. 65 (5) , 2432-2442
IRPCA: An Interpretable Robust Principal Component Analysis Framework for Inferring miRNA–Drug Associations
Recent evidence indicates that microribonucleic acids (miRNAs) are crucial in modulating drug sensitivity by orchestrating the expression of genes involved in drug metabolism and its pharmacological effects. Existing predictive methods struggle to extract features related to miRNAs and drugs, often overlooking the significance of data noise and the limitations of using a single similarity measure. To address these limitations, we propose an interpretable robust principal component analysis framework (IRPCA). IRPCA enhances the robustness of the model by employing a nonconvex low-rank approximation, thereby offering greater flexibility. Interpretability is ensured by analyzing low-rank matrix decomposition, which clarifies how miRNAs interact with drugs. Gaussian interaction profile kernel (GIPK) similarities are introduced to compute integrated similarities between miRNAs and drugs, addressing the issue of the single similarity feature. IRPCA is subsequently utilized to extract pertinent features, and a fully connected neural network is employed to generate the ultimate prediction scores. To assess the efficacy of IRPCA, we implemented 5-fold cross-validation (CV), which outperformed other leading methods, achieving the highest area under the curve (AUC) value of 0.9653. Additionally, case studies provide additional evidence supporting the efficacy of IRPCA.
- 21 February 2025
- Published by American Chemical Society (ACS) in Journal of Chemical Information and Modeling
- Vol. 65 (5) , 2432-2442
IRPCA: An Interpretable Robust Principal Component Analysis Framework for Inferring miRNA–Drug Associations
Recent evidence indicates that microribonucleic acids (miRNAs) are crucial in modulating drug sensitivity by orchestrating the expression of genes involved in drug metabolism and its pharmacological effects. Existing predictive methods struggle to extract features related to miRNAs and drugs, often overlooking the significance of data noise and the limitations of using a single similarity measure. To address these limitations, we propose an interpretable robust principal component analysis framework (IRPCA). IRPCA enhances the robustness of the model by employing a nonconvex low-rank approximation, thereby offering greater flexibility. Interpretability is ensured by analyzing low-rank matrix decomposition, which clarifies how miRNAs interact with drugs. Gaussian interaction profile kernel (GIPK) similarities are introduced to compute integrated similarities between miRNAs and drugs, addressing the issue of the single similarity feature. IRPCA is subsequently utilized to extract pertinent features, and a fully connected neural network is employed to generate the ultimate prediction scores. To assess the efficacy of IRPCA, we implemented 5-fold cross-validation (CV), which outperformed other leading methods, achieving the highest area under the curve (AUC) value of 0.9653. Additionally, case studies provide additional evidence supporting the efficacy of IRPCA.
- 1 November 2024
- Published by American Chemical Society (ACS) in Journal of Chemical Information and Modeling
- Vol. 64 (22) , 8641-8654
TIWMFLP: Two-Tier Interactive Weighted Matrix Factorization and Label Propagation Based on Similarity Matrix Fusion for Drug-Disease Association Prediction
Accurately identifying new therapeutic uses for drugs is crucial for advancing pharmaceutical research and development. Matrix factorization is often used in association prediction due to its simplicity and high interpretability. However, existing matrix factorization models do not enable real-time interaction between molecular feature matrices and similarity matrices, nor do they consider the geometric structure of the matrices. Additionally, efficiently integrating multisource data remains a significant challenge. To address these issues, we propose a two-tier interactive weighted matrix factorization and label propagation model based on similarity matrix fusion (TIWMFLP) to assist in personalized treatment. First, we calculate the Gaussian and Laplace kernel similarities for drugs and diseases using known drug-disease associations. We then introduce a new multisource similarity fusion method, called similarity matrix fusion (SMF), to integrate these drug/disease similarities. SMF not only considers the different contributions represented by each neighbor but also incorporates drug-disease association information to enhance the contextual topological relationships and potential features of each drug/disease node in the network. Second, we innovatively developed a two-tier interactive weighted matrix factorization (TIWMF) method to process three biological networks. This method realizes for the first time the real-time interaction between the drug/disease feature matrix and its similarity matrix, allowing for a better capture of the complex relationships between drugs and diseases. Additionally, the weighted matrix of the drug/disease similarity matrix is introduced to preserve the underlying structure of the similarity matrix. Finally, the label propagation algorithm makes predictions based on the three updated biological networks. Experimental outcomes reveal that TIWMFLP consistently surpasses state-of-the-art models on four drug-disease data sets, two small molecule-miRNA data sets, and one miRNA-disease data set.
- 1 November 2024
- Published by American Chemical Society (ACS) in Journal of Chemical Information and Modeling
- Vol. 64 (22) , 8641-8654
TIWMFLP: Two-Tier Interactive Weighted Matrix Factorization and Label Propagation Based on Similarity Matrix Fusion for Drug-Disease Association Prediction
Accurately identifying new therapeutic uses for drugs is crucial for advancing pharmaceutical research and development. Matrix factorization is often used in association prediction due to its simplicity and high interpretability. However, existing matrix factorization models do not enable real-time interaction between molecular feature matrices and similarity matrices, nor do they consider the geometric structure of the matrices. Additionally, efficiently integrating multisource data remains a significant challenge. To address these issues, we propose a two-tier interactive weighted matrix factorization and label propagation model based on similarity matrix fusion (TIWMFLP) to assist in personalized treatment. First, we calculate the Gaussian and Laplace kernel similarities for drugs and diseases using known drug-disease associations. We then introduce a new multisource similarity fusion method, called similarity matrix fusion (SMF), to integrate these drug/disease similarities. SMF not only considers the different contributions represented by each neighbor but also incorporates drug-disease association information to enhance the contextual topological relationships and potential features of each drug/disease node in the network. Second, we innovatively developed a two-tier interactive weighted matrix factorization (TIWMF) method to process three biological networks. This method realizes for the first time the real-time interaction between the drug/disease feature matrix and its similarity matrix, allowing for a better capture of the complex relationships between drugs and diseases. Additionally, the weighted matrix of the drug/disease similarity matrix is introduced to preserve the underlying structure of the similarity matrix. Finally, the label propagation algorithm makes predictions based on the three updated biological networks. Experimental outcomes reveal that TIWMFLP consistently surpasses state-of-the-art models on four drug-disease data sets, two small molecule-miRNA data sets, and one miRNA-disease data set.
- 24 July 2023
- Published by Oxford University Press (OUP) in Briefings in Bioinformatics
- Vol. 24 (5)
Generative Adversarial Matrix Completion Network based on Multi-Source Data Fusion for miRNA–Disease Associations Prediction
Numerous biological studies have shown that considering disease-associated micro RNAs (miRNAs) as potential biomarkers or therapeutic targets offers new avenues for the diagnosis of complex diseases. Computational methods have gradually been introduced to reveal disease-related miRNAs. Considering that previous models have not fused sufficiently diverse similarities, that their inappropriate fusion methods may lead to poor quality of the comprehensive similarity network and that their results are often limited by insufficiently known associations, we propose a computational model called Generative Adversarial Matrix Completion Network based on Multi-source Data Fusion (GAMCNMDF) for miRNA–disease association prediction. We create a diverse network connecting miRNAs and diseases, which is then represented using a matrix. The main task of GAMCNMDF is to complete the matrix and obtain the predicted results. The main innovations of GAMCNMDF are reflected in two aspects: GAMCNMDF integrates diverse data sources and employs a nonlinear fusion approach to update the similarity networks of miRNAs and diseases. Also, some additional information is provided to GAMCNMDF in the form of a ‘hint’ so that GAMCNMDF can work successfully even when complete data are not available. Compared with other methods, the outcomes of 10-fold cross-validation on two distinct databases validate the superior performance of GAMCNMDF with statistically significant results. It is worth mentioning that we apply GAMCNMDF in the identification of underlying small molecule-related miRNAs, yielding outstanding performance results in this specific domain. In addition, two case studies about two important neoplasms show that GAMCNMDF is a promising prediction method.
- 24 July 2023
- Published by Oxford University Press (OUP) in Briefings in Bioinformatics
- Vol. 24 (5)
Generative Adversarial Matrix Completion Network based on Multi-Source Data Fusion for miRNA–Disease Associations Prediction
Numerous biological studies have shown that considering disease-associated micro RNAs (miRNAs) as potential biomarkers or therapeutic targets offers new avenues for the diagnosis of complex diseases. Computational methods have gradually been introduced to reveal disease-related miRNAs. Considering that previous models have not fused sufficiently diverse similarities, that their inappropriate fusion methods may lead to poor quality of the comprehensive similarity network and that their results are often limited by insufficiently known associations, we propose a computational model called Generative Adversarial Matrix Completion Network based on Multi-source Data Fusion (GAMCNMDF) for miRNA–disease association prediction. We create a diverse network connecting miRNAs and diseases, which is then represented using a matrix. The main task of GAMCNMDF is to complete the matrix and obtain the predicted results. The main innovations of GAMCNMDF are reflected in two aspects: GAMCNMDF integrates diverse data sources and employs a nonlinear fusion approach to update the similarity networks of miRNAs and diseases. Also, some additional information is provided to GAMCNMDF in the form of a ‘hint’ so that GAMCNMDF can work successfully even when complete data are not available. Compared with other methods, the outcomes of 10-fold cross-validation on two distinct databases validate the superior performance of GAMCNMDF with statistically significant results. It is worth mentioning that we apply GAMCNMDF in the identification of underlying small molecule-related miRNAs, yielding outstanding performance results in this specific domain. In addition, two case studies about two important neoplasms show that GAMCNMDF is a promising prediction method.