Nucleotide sequence and DNaseI sensitivity are predictive of 3D chromatin architecture

Abstract

Motivation: Recently, Hi-C has been used to probe the 3D chromatin architecture of multiple organisms and cell types. The resulting collections of pairwise contacts across the genome have connected chromatin architecture to many cellular phenomena, including replication timing and gene regulation. However, high resolution (10 kb or finer) contact maps remain scarce due to the expense and time required for collection. A computational method for predicting pairwise contacts without the need to run a Hi-C experiment would be invaluable in understanding the role that 3D chromatin architecture plays in genome biology. Results: We describe Rambutan, a deep convolutional neural network that predicts Hi-C contacts at 1 kb resolution using nucleotide sequence and DNaseI assay signal as inputs. Specifically, Rambutan identifies locus pairs that engage in high confidence contacts according to Fit-Hi-C, a previously described method for assigning statistical confidence estimates to Hi-C contacts. We first demonstrate Rambutan's performance across chromosomes at 1 kb resolution in the GM12878 cell line. Subsequently, we measure Rambutan's performance across six cell types. In this setting, the model achieves an area under the receiver operating characteristic curve between 0.7662 and 0.8246 and an area under the precision-recall curve between 0.3737 and 0.9008. We further demonstrate that the predicted contacts exhibit expected trends relative to histone modification ChIP-seq data, replication timing measurements, and annotations of functional elements such as promoters and enhancers. Finally, we predict Hi-C contacts for 53 human cell types and show that the predictions cluster by cellular function. Availability: Tutorials and source code for Rambutan are publicly available at https://github.com/ jmschrei/rambutan.

Nucleosome repositioning during differentiation of a human myeloid leukemia cell line

“…Our observation that, during HL-60/S4 differentiation, nucleosome loss and nucleosome gain affect gene expression asymmetrically (i.e., a clear association with nucleosome loss and a more complicated response to nucleosome gain) agrees with the concept that there are several different mechanisms by which nucleosome positioning can affect gene expression. Interestingly, these changes in nucleosome occupancy involve extended regions of one to several kb, which is different from previously considered single-nucleosome repositioning scenarios and parallels a similar effect recently found at the V(D)J recombination domains 48 The molecular mechanisms of the change of chromatin density in extended regions that we report here may be linked to chromatin compartmentalization in domains such as TADs (topologically associated domains), 49,50 which have been recently shown to be correlated with regions of differential chromatin accessibility 51 . An exploration of this idea may provide a direction for new studies.…”

Computational Chromosome Conformation Capture by Correlation of ChIP-seq at CTCF motifs

“…However, these approaches lose the tissue specificity of the interactions. Other approaches integrate many diverse chromatin signals such as posttranslational histone modifications, chromatin accessibility, or transcriptional activity [57][58][59][60][61][62], and combine them with sequence features [63], or evolutionary constrains [64]. While these methods predict enhancer-gene association with good performance, they require for each specific condition of interest a multiplicity of input datasets, which are often not available.…”

Graph Convolutional Networks for Epigenetic State Prediction Using Both Sequence and 3D Genome Data

“…[34] identified coupling DNA motif pairs on long-range chromatin interactions. [25] use convolutional neural networks to predict Hi-C interactions from sequence inputs. None of the previous methods, however, use known Hi-C data to learn better feature representations of genomic sequences for epigenetic state prediction.…”