What to do when a bug is found in a BioConductor package (Glimma)
1
0
Entering edit mode
Sam ▴ 10
@sam-21502
Last seen 7 weeks ago
Jerusalem

As far as I can tell, I have found a bug in Glimma.

The following code

library(Glimma)

DE <- as.data.frame(cbind(202197.7, -1.410116), byrow=T)

colnames(DE) <- c("baseMean", "log2FoldChange")
DE$sig <- 1 glMDPlot(x=DE, xval="baseMean", yval="log2FoldChange", counts=NULL, anno=NULL,samples=NULL, group = NULL, status=DE$sig,
side.main = "baseMean",html="One_Gene_from_vector_final",
display.columns = c("baseMean","log2FoldChange"))


produces the following graph :

As you can see, the baseMean is rounded (incorrectly) to 202200.

What is the etiquette in this case? Should I email the developers directly?

glimma software error • 342 views
1
Entering edit mode

Thanks for the bug report Samuel. As Gordon mentioned, I would prefer bugs be submitted through Github issues. The reason for the bug is that I've rounded all values to 4 significant places to avoid values with tens of digits, it's clear that this causes confusion for large numeric values so I'll update the rounding strategy soon.

1
Entering edit mode
@gordon-smyth
Last seen 2 hours ago
WEHI, Melbourne, Australia

If you type packageDescription("Glimma"), or visit the Glimma landing page at the http://bioconductor.org/packages/Glimma, you will see that Glimma has an issues site for reporting bugs or other issues:

https://github.com/Shians/Glimma/issues

I'm not sure that rounding would be considered a bug though.