Correlation between accuracy and Sig scores. Non-degenerate 6-mers from S. cerevisiae were scored according to Sig scores based on (a) Overrepresentation, (b) Overrepresentation-KS, (c) Coverage and (d) KS metrics of statistical significance. The 6-mers were randomly sampled from both the upstream regions and the known binding sites to ensure coverage or a wide range of accuracy. The x-axis plots the Bonferroni-corrected and log2 transformed Sig score for each metric. The red lines indicate the 95th Sig percentile.
Chakravarty et al. BMC Bioinformatics 2007 8:249 doi:10.1186/1471-2105-8-249