Fig. 1

Overview of interaction score thresholding procedure for cell type-specific looping interaction classification. a A 5C dataset is input as a set of interaction frequency matrices, with each matrix capturing the same set of genomic contacts under a different cellular condition. b Raw 5C counts are converted to interaction scores (IS) which reflect bias-corrected, sequencing depth normalized, local expected background signal normalized, and statistically modeled interaction frequency values that are comparable within and between conditions under the assumptions of our model (detailed in the “Methods” section and Fig. 4). c Interaction scores are thresholded to allow detection and classification of looping interactions that are significantly differential across cellular conditions. d Seven looping interaction classes after a 3-way thresholding scheme on ES-2i, ES-serum, and NPC cellular states. e IS heatmaps at two selected genomic loci. Green boxes highlight regions of qualitatively apparent differences in looping signal. f Loop classification results after applying 3DeFDR-5C’s 3-way IS thresholding procedure