CharPlant

CharPlant predicts chromatin accessible regions (OCRs) de novo across plant genomes to identify putative open chromatin for studying transcriptional regulation.

Key Features:

• De Novo Prediction: Predicts OCRs directly from DNA sequence, enabling genome-wide identification of putative open chromatin without relying solely on experimental accessibility assays.
• Three-Layer Convolutional Neural Network: Implements a three-layer CNN trained on DNase-seq and ATAC-seq datasets from four plant species.
• Motif and Regulatory Pattern Learning: Learns sequence motifs and regulatory patterns that govern DNA accessibility to inform predictions.
• Output Format: Outputs predicted OCR coordinates in .Bed format.

Scientific Applications:

• Transcriptional Regulation Studies: Map putative OCRs to study gene transcription regulation across different tissues and developmental stages.
• Stress and Stimulus Response Analysis: Explore dynamic chromatin accessibility associated with stress responses and environmental stimuli.
• Developmental Transition Investigations: Characterize chromatin changes during developmental transitions to inform plant biology and agricultural biotechnology research.

Methodology:

A three-layer convolutional neural network was constructed and trained on DNase-seq and ATAC-seq datasets from four plant species to learn sequence motifs and regulatory patterns for genome-wide OCR prediction.