deepC

deepC predicts Hi-C chromatin interactions from megabase-scale DNA sequences to model three-dimensional genome folding and to assess effects of genetic variation on chromatin architecture.

Key Features:

• Hi-C interaction prediction: Predicts Hi-C chromatin interactions from megabase-scale DNA sequences.
• Transfer-learning-based deep learning: Implements a transfer-learning-based deep neural network framework to learn sequence determinants of genome folding.
• Domain boundary identification: Identifies domain boundaries at high resolution from sequence-based predictions.
• Variant impact prediction: Predicts the impact of large-scale structural variations and single base-pair changes on three-dimensional genome architecture.
• Noncoding variation interpretation: Interprets noncoding genetic variation in the context of three-dimensional genome organization.

Scientific Applications:

• Noncoding genetic variation analysis: Assess functional consequences of noncoding genetic variation using predicted three-dimensional genome structure.
• Structural genomics: Study effects of structural variants on chromatin folding and domain boundaries.
• Gene regulation research: Investigate mechanisms of gene regulation driven by three-dimensional genome organization.
• Genetic variation studies: Analyze how single base-pair changes and larger variants influence chromosomal organization and regulatory architecture.

Methodology:

Uses a transfer-learning-based deep neural network trained on megabase-scale DNA sequences to predict Hi-C chromatin interactions and domain boundaries and to evaluate effects of structural variations and single base-pair changes.