The 11th International Symposium on 3D Genomics

110 / 2024-10-30 13:22:38

G-3D: A Deep Learning Framework for Predicting 3D Genome Architecture Across Cell Types Using Transcription Factor Expression

Transcription factors (TFs),Deep learning,3D genome organization

Transcription factor(TF) binding on chromatin impacts the recruitment of cellular machinery needed for gene transcription and histone modification, thereby affecting cell type-specific chromatin 3D structures and gene expression. Numerous studies have shown that knockdown or overexpression of individual transcription factors can influence chromatin 3D structure. However, no systematic research has yet evaluated the extent to which different transcription factor expressions impact specific DNA sequences. To achieve this goal, we utilize cross-attention to integrate DNA sequences and transcription factor expressions to predict cell type-specific chromatin 3D structures. Through this method, we identified transcription factors with significant impacts on chromatin 3D structure. We aim, to map the relationship between the expression levels of these transcription factors and the chromatin 3D structures formed by specific DNA sequences, highlighting how different transcription factors impact distinct DNA regions. We hope this model will provide insights for manipulating chromatin 3D structures artificially.