Skip to main content
Fig. 2 | Genome Biology

Fig. 2

From: 3DeFDR: statistical methods for identifying cell type-specific looping interactions in 5C and Hi-C data

Fig. 2

5C counts are overdispersed and their mean-variance relationship varies as a function of linear genomic distance and cellular condition. a Raw 5C contacts are stratified by genomic distance prior to characterization of their mean-variance relationship. In each of our three regimes, the width of the stratification windows is determined using a different binning scheme. b The coefficient of variation for raw 5C counts is plotted against the median genomic interaction distance for each sliding window. Each window captures counts from all genomic regions in the dataset in the ES-2i condition. c The dispersion parameter, A, for each distance scale window (short horizontal lines) is computed by fitting sample means and variances to the function σ2 = A*μ2 + μ. Dispersion versus distance scale trends (solid smooth lines) were generated by Loess smoothing. d Mean-variance models for representative genomic distance windows from all three distance regimes. Fits of the Poisson mean-variance relationship (σ2 = μ) and the negative binomial mean-variance relationship (σ2 = A*μ2 + μ) are shown with their corresponding R2 goodness of fit values

Back to article page