DeepCMI

DeepCMI predicts circRNA–miRNA interactions (CMIs) using a graph-based model that integrates multi-source circRNA and miRNA data to identify potential CMIs involved in gene regulation via competing endogenous RNA mechanisms.

Key Features:

• Graph-Based Modeling: DeepCMI employs a graph-based approach to integrate diverse circRNA and miRNA data for interaction prediction.
• Multi-Source Information Utilization: The model leverages comprehensive datasets encompassing known interactions and biological characteristics.
• High Predictive Accuracy: DeepCMI achieved Area Under the Curve (AUC) values of 90.54% on the CMI-9905 dataset and 94.8% on the CMI-9589 dataset.
• Reduction of In Vitro Study Requirements: Accurate in silico predictions decrease reliance on resource-intensive in vitro experiments for identifying candidate CMIs.

Scientific Applications:

• Gene Regulation Analysis: Predicting CMIs to study circRNA- and miRNA-mediated gene regulation via competing endogenous RNA mechanisms.
• Molecular and Genetic Research: Supporting studies in molecular biology and genetics that require interaction networks between circRNAs and miRNAs.
• Functional Studies of Reproduction and Apoptosis: Facilitating investigation of CMIs implicated in reproduction and apoptosis.

Methodology:

DeepCMI constructs a graph-based computational model that integrates multiple circRNA and miRNA data types and known interaction information to predict circRNA–miRNA interactions.