Multi-Relation Graph Embedding for Predicting miRNA-Target Gene Interactions by Integrating Gene Sequence...

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• MicroRNAs (miRNAs) play a critical role in complex diseases like cancer by modulating gene expression.
• Accurate prediction of miRNA-target interactions is essential for understanding miRNA functions and disease mechanisms.
• Large-scale biological data and heterogeneous networks offer new avenues for miRNA target identification.

Purpose of the Study:

• To develop an efficient computational method for predicting miRNA-target interactions.
• To improve prediction accuracy by integrating network topology and sequential gene information.
• To propose a novel graph-based model, MRMTI, for enhanced miRNA-target interaction prediction.

Main Methods:

• Proposed a graph-based model named MRMTI.
• Utilized a multi-relation graph convolution module to capture network topology.
• Employed a Bi-LSTM module to incorporate sequential gene information.
• Calculated prediction scores using learned miRNA and gene embeddings.

Main Results:

• MRMTI demonstrated superior performance compared to state-of-the-art graph embedding methods and existing bioinformatic tools.
• Experimental results showed the effectiveness of multi-relation learning through MRMTI variants.
• Case studies confirmed MRMTI's capability in predicting novel miRNA-target associations.

Conclusions:

• MRMTI is a highly effective computational tool for predicting miRNA-target interactions.
• The integration of network and sequential data significantly enhances prediction accuracy.
• MRMTI holds promise for advancing research in miRNA-mediated gene regulation and disease studies.