Closed kevinrue closed 10 years ago
fixed by commit a3f488d98ab8bc77be5328d34106bf6fdf355e0b
The second round of anova calculations, which was getting the p-value, is now skipped. This is more conistent with the randomForest approach which also keeps the statistic of all genes, without measure of significance. Moreover, non-significant genes in the anova framework do not affect much high-ranking GO terms anyway
The double ANOVA per gene can surely be avoided. This could reduce by half the duration of the analysis.