Enhancing Disease Prediction through lncRNA Analysis: A Multi-Network Biomolecular Approach

Jiongmin Zhang
Computer Science and Technology, East China Normal University, Shanghai, China
Weijun Chen
Computer Science and Technology, East China Normal University, Shanghai, China
Ying Qian
Computer Science and Technology, East China Normal University, Shanghai, China

DOI:https://doi.org/10.5912/jcb1433

Abstract:

The relationship between long non-coding RNAs (lncRNAs) and various human diseases has become a focal point in biotechnological research due to their significant biological roles. However, the volume of experimentally validated associations between lncRNAs and diseases remains limited, underscoring the need for predictive methodologies to uncover potential lncRNA-disease connections. This paper introduces a novel computational model, termed LMPDN-LDA, designed to predict lncRNA-disease associations (LDAs) by leveraging multiple biomolecular networks involving lncRNAs, miRNAs, proteins, and diseases. The LMPDN-LDA model comprises three primary components: First, a comprehensive association network is constructed by integrating known interaction data among diseases, proteins, miRNAs, and lncRNAs. Subsequently, a graph embedding technique is employed to learn vector representations of all entities within this network. Finally, these vectorized features are utilized within a Random Forest (RF) classifier to predict novel LDAs. Empirical validation through cross-validation and targeted case studies demonstrates that the LMPDN-LDA model effectively identifies potential lncRNA-disease associations, significantly advancing our capability to predict disease-related biomolecular interactions. This model not only enhances predictive accuracy but also contributes novel insights into the mechanisms underlying lncRNA involvement in diseases, offering valuable pathways for future biotechnological research and therapeutic development.