EDLMFC

EDLMFC predicts interactions between non-coding RNAs (ncRNAs) and proteins by integrating multi-scale sequence and structural features for accurate interaction inference.

Key Features:

• Multi-Scale Feature Integration: Integrates primary sequence features, secondary structure sequence features, and tertiary structure features to improve prediction accuracy.
• Conjoint k-mer Extraction: Employs conjoint k-mer techniques to extract comprehensive protein and ncRNA sequence features and integrates them with tertiary structural information.
• Ensemble Deep Learning Model: Uses an ensemble deep learning approach combining a convolutional neural network (CNN) to identify dominant biological patterns and a bi-directional long short-term memory network (BLSTM) to capture long-range dependencies.

Scientific Applications:

• Predictive Accuracy: Reports accuracies of 93.8%, 89.7%, and 86.1% on the RPI1807, NPInter v2.0, and RPI488 datasets, respectively.
• Cross-Species Prediction: Demonstrates effectiveness in predicting potential ncRNA-protein interactions across different organisms as evidenced by independent testing.
• Network Analysis: Identifies hub ncRNAs and proteins within Mus musculus ncRNA-protein interaction networks to reveal network dynamics and key regulatory elements.

Methodology:

Performs conjoint k-mer extraction of protein and ncRNA sequences, extracts primary sequence, secondary structure sequence, and tertiary structure features, integrates these multi-scale features, and applies an ensemble deep learning framework combining CNN and BLSTM.