add.bar does not work as expected

[alexyfyf]

Hi Jun,

Thanks for developing this great package. However, when I use it, I notice some issues. I use add.bar to add a bar plot for GO, which I thought would be -log10pvalue. However, it turns out it uses the ratio rather than the p-value. Using ratio is fine, but the default GO label is ordered by p-value, so creating inconsistency, and moreover, the GO barplot only have a maximum of 4 bars even when I supply the enrichment table with 5 GO terms for each cluster.

Here is a screenshot of a cluster You can see the bar plot and GO term labels are not ordered in the same way, and missing 1 term as well. You can also see the enrichment table for this cluster is like this.

And this is the code I used to plot the figure:

visCluster(object = gene_ck10,
           plot.type = "both",
           annoTerm.data = enrich_ck10,
           add.box = T, show_row_dend = F, add.bar = T,
           line.side = "left",
           markGenes.side = "left",
           boxcol = rep(hue_pal()(4)[1:2], each = 4),
           sample.col	= rep(c(hue_pal()(4)[c(1,4,3)], "#808080"), 2))

Could you please have a look if this is a bug, or I have done something wrong?

Cheers,

[junjunlab]

Please see ClusterGVis 对指定 cluster 添加注释.

[alexyfyf]

Thanks for showing this. Yes I can understand it uses the 4th column for the bar plot, and scales between 0 to 1. But it does look misleading, because 0 is not shown as a bar.
Also, I would assume the order of GO terms and bars to be the same at least, but it is not the case now.

[junjunlab]

And the GO table is arranged by pvalue instead of ratio. And some bugs I will fix, thanks for your advice.