Read density bias. This figure shows CHANCE output for the read density bias estimation module. The x-axis denotes length scale, from 1 kbp to 16.384 Mbp. The y-axis denotes percentage of variance in data at a given length scale (spectral energy). The blue curve is the user's data, while the red curve denotes the distribution of an idealized bias-free data set generated by a Poisson simulation (see Materials and methods). Noticeable deviation of the blue curve from the red simulation curve might indicate copy number alterations in cancer cells, amplification bias in generating duplicate reads, GC content and mappability bias, or inability to sonicate heterochromatin regions. (a) Mouse NSC Input from the Lim lab with significant IP enrichment, but also with a heavy amplification bias in the input channel. (b) After de-duplicating reads, the sample in (a) shows a reduction in bias as demonstrated by a better agreement between the user's data and the Poisson simulation (compare Figures 1b,c and 2b,c). (c) For comparison, this panel shows mouse whole limb Input from the Ahituv lab, which demonstrates relatively low bias in read density.
Diaz et al. Genome Biology 2012 13:R98 doi:10.1186/gb-2012-13-10-r98