Fig. 8
From: Identifying peaks in *-seq data using shape information
Outlier detection. The plot shows the Mahalanobis distance of each candidate region to the set of positive and negative profiles. The threshold values for positive and negative samples are indicated by a vertical and a horizontal line, respectively. Inner figures a, b, c, and d show examples of profiles, displaying reads matching to ChIP-seq data (top part, light blue background) and reads matching to the genomic control (bottom part, white background): a A noisy outlier is excluded from the set of positive profiles; b A very clear peak is classified as positive profile; c A peak with a high coverage in the control is classified as negative profile; d A region with abnormally high coverage is excluded from the set of negative profiles