Welcome to the MapMan Family of Software Forum

Please do not hesitate to register and post your question.

Don't forget to subscribe to your posted message so you get notified on updates.
Every question you post will help others and or enhance the software!

Post a question,   post a bug!

Welcome to the MapMen Family of Software Forum Welcome to the MapMen Family of Software Forum

MapMan Bugs

Fold change_infinity in RNA seq experiments

We are using MapMan to visualize RNA-seq results. in the case, when one gene is not expressed at all in one condition but has changed significantly, its fold change is like infinity but its p-value is <0.05, so it is significant. But it seems that MapMan can not handle this situation. it isvery common in RNA-seq experiments, and Cuffdiff also reports fold change like infinity, so there should be a way to handle this problem

RE: Fold change_infinity in RNA seq experiments
2/11/13 11:15 AM as a reply to Delasa Aghamirzaie.
Hi Delasa Aghamirzaie,

while the situation you describe is correct from a mathematical point of view is is unlikely
that, in the underlying biological system, the expression of the gene is really completely
zero in the sample that has zero counts. The abundance of the corresponding transcript
is simply too low to make sure it is picked up at the sensitivity level of your sequencing

Hence, a positive or negative infinity log2 fold change is a rather artificial value and the
information whether that was based on a 0 vs 40 or a 0 vs. 20000 counts is lost and
thereby the "impact" of such values appears exaggerated in the dataset.

A very simple way around this would be to simply set all "0" values to "1". A more
sophisticated approach would be to determine the noise margin in the data set and
set the "0" values to a value at or below this noise margin.

Subsequently the log2 fold change values can be recalculated to yield biologically
more meaningful values.

If we would include special handling of +/- Infinity values into MapMan (e.g. by assigning
a special color code to such values) we would still face the problem that +inf based on 0 vs 10
counts has a different biological significance than +inf based on 0 vs 20000 counts.

I hope this helps you tackle the problem - please don't hesitate to post again if you
have further questions or suggestions concerning this problem.