LPIH2V: LncRNA-protein interactions prediction using HIN2Vec based on heterogeneous networks model

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Long non-coding RNA (lncRNA)-protein interactions are integral to human disease development and treatment.
• Experimental methods for identifying lncRNA-protein interactions are costly and time-intensive.
• Existing computational prediction methods for lncRNA-protein interactions are limited.

Purpose of the Study:

• To develop an efficient and accurate computational model for predicting lncRNA-protein interactions.
• To address the urgent need for improved methods in this field.

Main Methods:

• Proposed LPIH2V, a heterogeneous network embedding model utilizing meta-paths.
• Constructed a heterogeneous network integrating lncRNA similarity, protein similarity, and known lncRNA-protein interaction networks.
• Employed the HIN2Vec method for extracting behavioral features within the heterogeneous network.

Main Results:

• LPIH2V achieved an Area Under the Curve (AUC) of 0.97 and an Accuracy (ACC) of 0.95 in 5-fold cross-validation.
• The model demonstrated superior performance and strong generalization capabilities compared to existing methods.
• LPIH2V effectively extracts both attribute (similarity) and behavioral (meta-path wandering) features.

Conclusions:

• LPIH2V offers a powerful and efficient approach for predicting lncRNA-protein interactions.
• The model's ability to integrate diverse network information enhances prediction accuracy.
• LPIH2V holds significant potential for advancing research in lncRNA-protein interactions and disease mechanisms.