Figure 1.

The colon cancer anti-profile signature. (A) Normalized gene expression for 15 hyper-variable genes in cancer from two independent colon cancer datasets [38-40]. Normal samples are shown in green, cancer samples are shown in orange. We define the anti-profile as the set of genes and a corresponding range of normal expression values for each gene (indicated by dotted lines). Only genes inside colon methylation blocks [23] were included. The anti-profile score for each sample is the number of genes in the signature that are outside their defined range of expression. Blue circles highlight expression for one specific cancer sample with an anti-profile score of 9. (B) ROC curves using the anti-profile method trained on one colon cancer study to score samples from an independent colon cancer study. The anti-profile includes genes inside colon DNA methylation change blocks where across-sample variance in cancer is at least twice that of normal in the training study. The anti-profile method is very accurate (ROCs of 0.94 and 1.00). (C) We compare the upper bounds of normal expression (median + 5*median absolute deviation) as defined by the two independent colon cancer studies and find that ranges are highly consistent.

Corrada Bravo et al. BMC Bioinformatics 2012 13:272   doi:10.1186/1471-2105-13-272
Download authors' original image