RLBind

RLBind predicts RNA–small molecule binding sites using a deep learning framework that integrates sequence-dependent and structure-dependent features. It applies a convolutional neural network (CNN) to capture global full-length RNA information and local nucleotide context for accurate binding site identification.

Key Features:

• Dual-Channel CNN Architecture: Combines global sequence-level features with local neighboring nucleotide information for binding site prediction.
• Sequence and Structure Feature Integration: Incorporates sequence-dependent and structure-dependent properties to improve prediction accuracy without requiring experimental tertiary structures.

Scientific Applications:

• RNA-Targeted Drug Discovery: Identifies RNA–small molecule interaction sites to support therapeutic targeting when experimental RNA structures are unavailable.

Methodology:

RLBind employs a convolutional neural network trained on RNA sequence and structural features, integrating full-length global context with local nucleotide patterns to predict RNA–ligand binding sites.