I would like to know if it's possible to distinguish between up and down regulated feature after running the stattest function in the ballgown package. In the github page refers to this "For two- and multi-group comparisons, a significant result indicates that the feature is differentially expressed in at least one of the groups", but i'm interested to know in which of the groups the feature is over or under expressed. The displayed fold change values are always positive, so i'm wondering if it's possible to extract this kind of information with this package.
Furthermore, FDR values seem weird, as in all the features with pvalue < 0.05 there isn't any with qvalue < 0.05 too. Why does this happen? Does it has something to do with the statistical model employed in ballgown?
Number of significant features:
> sum(bgresults_c1c3$pval < 0.05, na.rm = TRUE)  2759
Subset with signigicant features:
> subset <- subset(bgresults_c1c3, bgresults_c1c3$pval<0.05, na.rm=TRUE)
Histogram of qvalues:
> hist(subset$qval, main='Ballgown q-values: c1-c3 comparison', col="grey", xlab='Range of q-values for the more significant transcripts')
Thanks in advance,