iDeepMV

iDeepMV predicts RNA-binding protein (RBP) interactions with RNAs using multi-view deep learning and multi-label learning to capture binding similarities and correlations among RBPs.

Key Features:

• Multi-View Deep Feature Learning: Extracts multiple views including amino acid sequences and dipeptide components derived from RNA sequences for comprehensive feature representation.
• Integration with Multi-Label Learning and RRBN: Employs a multi-label learning framework supported by the RNA-RBP Binding Network (RRBN) to capture relationships and correlations between RBPs and multiple target RNAs.
• Deep Neural Networks for Feature Extraction: Uses deep neural network models tailored to each view to perform robust feature learning.
• Multi-Label Classifiers Across Three Views: Trains multi-label classifiers with interaction information across three views.
• Voting Mechanism for Decision Making: Implements a voting mechanism to integrate outputs from the multi-label classifiers for final predictions.

Scientific Applications:

• RBP–RNA Interaction Prediction: Improves accuracy of predicting RNA-RBP interactions relative to existing state-of-the-art methods.
• Gene Expression Regulation Studies: Facilitates analysis of RBP regulatory roles in gene expression regulation.
• Disease Mechanism Analysis: Supports study of disease mechanisms involving aberrant RBP activity.
• Therapeutic Target Identification: Assists identification of potential therapeutic targets related to RBP dysregulation.

Methodology:

Extracts multi-view data from RNA sequences (including amino acid sequences and dipeptide components); applies deep neural networks per view for feature learning; trains multi-label classifiers with interaction information across three views; and integrates classifier outputs using a voting mechanism.