PrismNet

PrismNet predicts cell type-specific RNA-binding protein (RBP)–RNA interactions by integrating in vivo RNA structural data with RBP binding information to model how RNA structure influences binding.

Key Features:

• Integration of in vivo data: Integrates RNA secondary-structure profiles from icSHAPE experiments with RBP binding-site information derived from UV cross-linking and immunoprecipitation within the same cell lines.
• Deep learning framework: Applies deep learning to process both sequential and structural RNA data for prediction of RBP binding likelihoods.
• Binding probability scores: Produces quantitative binding probability scores for RBP–RNA interactions.
• Saliency maps: Generates saliency maps that highlight sequence and structural regions contributing to predicted binding.
• Sequence-structure integrative motif identification: Identifies motifs that combine sequence and structural features associated with RBP binding.

Scientific Applications:

• Post-transcriptional regulation studies: Enables analysis of RBP–RNA interactions relevant to post-transcriptional control of gene expression.
• Elucidation of regulatory mechanisms: Supports investigation of how RNA structure and RBP binding jointly govern gene-expression regulation in specific cell types.
• Therapeutic strategy research: Informs development of targeted therapeutic approaches by identifying cell type-specific RBP–RNA interaction determinants.

Methodology:

Uses a deep learning framework that integrates in vivo icSHAPE-derived RNA structural profiles with RBP binding-site data from UV cross-linking and immunoprecipitation to predict binding probabilities, generate saliency maps, and identify sequence-structure motifs.