Comprehensive in silico functional characterization of non-coding variants via machine learning-based chromatin organization prediction
The mammalian genome is spatially organized in the nucleus to enable cell type-specific gene expression. The organization of the genome frequently goes awry in diseases when mutations - mostly non-coding ones - disrupt the underlying regulatory mechanism. We recently developed a scalable deep neural network model, C.Origami, that performs de novo prediction of cell type-specific chromatin organization at high resolution. This model enables in silico genetic experiments to examine the impact of genetic perturbations on chromatin interactions. We aim to leverage the C.Origami model to infer disease-causing mutations that disrupts chromatin interactions and gene expression, as well as their impacted cell types.