Fig. 3
From: Gene set analysis controlling for length bias in RNA-seq experiments
Proportion of DE of GO categories for unweighted and weighted SeqGSA. Genes in the LNCaP data set in [4] are mapped to the GO categories. The 1585 mapped GO categories are grouped by median gene length from low to high. For the unweighted method (left), the correlation of the proportion of DE categories and gene length is on the borderline of significance (p=0.068). For the weighted method (right), the correlation is not significant (p=0.117)