Cell-type-specific prediction of 3D chromatin organization enables high-throughput in silico genetic screening

Abstract

Investigating how chromatin organization determines cell-type-specific gene expression remains challenging. Experimental methods for measuring three-dimensional chromatin organization, such as Hi-C, are costly and have technical limitations, restricting their broad application particularly in high-throughput genetic perturbations. We present C.Origami, a multimodal deep neural network that performs de novo prediction of cell-type-specific chromatin organization using DNA sequence and two cell-type-specific genomic features—CTCF binding and chromatin accessibility. C.Origami enables in silico experiments to examine the impact of genetic changes on chromatin interactions. We further developed an in silico genetic screening approach to assess how individual DNA elements may contribute to chromatin organization and to identify putative cell-type-specific trans-acting regulators that collectively determine chromatin architecture. Applying this approach to leukemia cells and normal T cells, we demonstrate that cell-type-specific in silico genetic screening, enabled by C.Origami, can be used to systematically discover novel chromatin regulation circuits in both normal and disease-related biological systems.

Exploring the Roles of RNAs in Chromatin Architecture Using Deep Learning

“…4-6). This is consistent with results from models that incorporate epigenetic features such as CTCF binding or histone modifications 42 . Of the three RNA features, trans -located caRNA signals led to the…”

“…4-6). This is consistent with results from models that incorporate epigenetic features such as CTCF binding or histone modifications 42 . Of the three RNA features, trans-located caRNA signals led to the AkitaR model with the highest performance, closely followed by nascent RNA, and then steadystate transcription (Fig.…”

“…Since these models are highly accurate, they enable researchers to decode sequence determinants of genome folding through computational techniques such as in silico mutagenesis and feature importance scores 41 . More recently, models incorporating epigenomic data were built to achieve cell-type-specific predictions 42,43 . Significantly, these models learned the sequence and epigenetic correlates of 3D genome folding.…”

Considerations and caveats for analyzing chromatin compartments

“…In addition to these strategies focused on compartment prediction, we wish to note that there are several methods that are designed to predict signal de novo or to enhance low-depth signal within 2D contact maps ( Zhang et al, 2018 ; Carron et al, 2019 ; Liu et al, 2019 ; Liu and Wang, 2019 ; Schwessinger et al, 2020 ; Cheng et al, 2021 ; Tan et al, 2023 ). These methods often use neural networks and are now demonstrating remarkable accuracy.…”

“…It is important to consider how sequencing depth impacts our understanding of compartments as ultra-deep sequencing reveals many new aspects of fine-scale compartmental organization ( Gu et al, 2021 ). The increasing ability to use machine learning to impute high-resolution data from maps with low sequencing depth may also help in this regard ( Zhang et al, 2018 ; Carron et al, 2019 ; Liu et al, 2019 ; Liu and Wang, 2019 ; Schwessinger et al, 2020 ; Cheng et al, 2021 ; Tan et al, 2023 ). As the cost of sequencing decreases and deeply sequenced Hi-C and Micro-C maps become more common, it will be valuable determine the effectiveness of these imputation methods at fine-scale.…”