Detecting local changes in chromatin architecture with false discovery control
Koch, H.; Yang, T.; Imakaev, M.; Hardison, R. C.; Li, Q.
Show abstract
Hi-C experiments are a powerful means to describe the organization of chromatin interactions genome-wide. By using Hi-C data to identify differentially organized genomic regions, relationships between this organization, gene expression, and cell identity may be established. However, Hi-C data exhibit a unique and challenging spatial structure, as genomic loci can show strong correlations when they are nearby in 3D space within the nucleus or 1D space along the chromosome. Consequently, the development of methods that can accurately detect differences between Hi-C samples while controlling false discoveries has remained difficult. To meet this need, we introduce a spatial modeling approach based on sliding window statistics. Using polymer simulations, we illustrate the improved power and precision of our method to identify differentially interacting genomic regions. We further demonstrate our methods ability to reveal biologically meaningful changes in chromatin architecture through two data analyses concerning the loss of architectural and chromatin remodeling proteins.
Matching journals
The top 4 journals account for 50% of the predicted probability mass.
Similar papers in this journal
Similar papers in this journal
- Stripenn detects architectural stripes from chromatin conformation data using computer vision 96%
- Massively parallel reporter perturbation assay uncovers temporal regulatory architecture during neural differentiation 96%
- Landscape of allele-specific transcription factor binding in the human genome 96%
Similar papers in this journal
Similar papers in this journal
- Predicting gene expression from histone marks using chromatin deep learning models depends on histone mark function, regulatory distance and cellular states 95%
- Epitome: Predicting epigenetic events in novel cell types with multi-cell deep ensemble learning 95%
- Interpretable trajectory inference with single-cell Linear Adaptive Negative-binomial Expression (scLANE) testing 95%
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.