DNAcycP

DNAcycP predicts intrinsic DNA cyclizability from DNA sequence using a deep learning model trained on loop-seq assay data to quantify DNA bendability.

Key Features:

• Deep Learning Framework: Implements a deep learning model trained on loop-seq assay data to predict intrinsic DNA cyclizability.
• High Fidelity Predictions: Produces predictions validated against experimental loop-seq data.
• Cyclizability Score (C-score): Outputs a quantitative C-score that distinguishes DNA fragments by loopability.

Scientific Applications:

• Comparative Genomic Analysis: Applied to yeast and mouse genomes to reveal conserved high DNA bendability around nucleosome dyads.
• Transcription Factor Binding Sites: Identifies elevated cyclizability at CTCF binding sites in the mouse genome, a property conserved across mammals.
• Motif Analysis: Highlights mechanical properties associated with specific DNA motifs, particularly those involved in transcription factor binding.

Methodology:

Trains a deep learning model on loop-seq assay data and validates predictions using independent in vitro selection datasets that enrich for loopable sequences.