TriplexFPP

TriplexFPP predicts DNA:RNA triplex formation potential to identify triplex-forming long non-coding RNAs (lncRNAs) and candidate DNA triplex sites for studies of lncRNA-mediated genomic regulation.

Key Features:

• Integrated Prediction Models: Two models predict triplex-forming entities: a Triplex lncRNA Prediction Model that identifies lncRNAs capable of forming triplex structures with DNA, and a Triplex DNA Site Prediction Model that predicts potential DNA triplex sites.
• Machine Learning Approach: Convolutional neural networks (CNNs) learn high-level features from experimentally verified datasets to improve prediction accuracy.
• Performance Metrics: For triplex-forming lncRNAs (redundancy removal threshold 0.8) the average ROC AUC is 0.9649 and PRC AUC is 0.9996, and for triplex DNA site predictions the ROC AUC is 0.8705 and PRC AUC is 0.9671.

Scientific Applications:

• lncRNA functional inference: Predicts lncRNAs likely to form DNA:RNA triplexes to support analyses of lncRNA regulatory roles.
• Cis and trans targeting analysis: Identifies candidate triplex interactions relevant to cis and trans targeting mechanisms of triplex-forming lncRNAs.
• Genomic regulation studies: Provides candidate DNA sites and lncRNAs for experimental investigation of triplex-mediated genomic regulation.

Methodology:

Convolutional neural networks trained on experimentally verified datasets are used to learn high-level features for predicting DNA:RNA triplex formation potential.