The advantages of smoothing. (a) Points represent single-CpG methylation estimates plotted against their genomic location. Large points are based on greater than 20× coverage. The orange circle denotes the location for which we are estimating the methylation profile. The blue points are those receiving positive weight in the local likelihood estimation. The orange line is obtained from the fitted parabola. The black line is the methylation profile resulting from repeating the procedure for each location. (b) The curve represents the kernel used in the weighted regression and the points are the actual weights, which are also influenced by coverage. (c) Points are as in (a) for the 25× coverage Lister data. The pink line is obtained by applying BSmooth to a the full data. The black line is the estimate from BSmooth based on a 5× subset of the Lister data. (d) The points are as in (a) but for the Hansen-capture data with average 35× coverage, and average across three replicates. The black line is the BSmooth estimate obtained from the 4× Hansen data, averaged across three replicates.
Hansen et al. Genome Biology 2012 13:R83 doi:10.1186/gb-2012-13-10-r83