Skip to main content
. 2011 Jan 19;27(5):662–669. doi: 10.1093/bioinformatics/btr005

Fig. 5.

Fig. 5.

Comparison of different methods for adjusting transcript length bias based on the Fisher's exact test. Each point represents 65 GO terms with similar median transcript length. Significance level for GO terms is nominal P-value of 0.05. The significant gene list for microarray data is generated using significant level of FDR 0.025, while that for RNA-seq is generated using significant level of 0.05 with Bonferroni correction. Different significant levels are used to avoid extremely short gene list from microarray data.