|
Resolution: standard / high Figure 5.
Functional categories of genes can be predicted by differential coexpression. The blue line curve shows the result of leave one out validation to generate an
ROC curve for reconstructing GO categories containing 10-30 genes using their differential
coexpression values with AUC 0.81. The red line shows if coexpression values alone
are used (0.74). The black line curve shows the resulting ROC curve if in place of
GO sets, random sets with 10-30 genes are used (AUC 0.49).
Gillis and Pavlidis BMC Bioinformatics 2009 10:306 doi:10.1186/1471-2105-10-306 |