NormFinder Stability Value vs SD

Here we discuss aelection and normalization of reference genes

Moderator: MultiD Support

NormFinder Stability Value vs SD

Postby TL&Hoh » Fri Aug 19, 2016 3:38 pm

Dear MultiD support,

I have a (hopefully only) semantic question regarding the NormFinder output.

In the Anderson paper from 2004 on the NormFinder algorithm the authors define the "Stability Value". The output of NormFinder in GenEx is called SD, standard deviation. Now the Anderson paper does not really mention SD and on the other hand I cannot find a reference to the "Stability Value" anywhere in the GenEx documentation (help HTML).
Now while I think it is clear I understand how to get to the most stable gene using NormFinder in GenEx, I would still like to know what is the relationship between "Stability Value" and "SD". Are these the same, was something changed in the algorithm, or is there some additional calculation neccessary to get from "SD" to "Stability Value"?

Thank you

Posts: 1
Joined: Thu Aug 04, 2016 11:58 am

Re: NormFinder Stability Value vs SD

Postby Mikael Kubista » Fri Aug 19, 2016 5:09 pm

Hi Tobias,
You find the relation between "stability values" and "SD" in the Supplementary material to the Andersens paper. "Stability values" in the Anderson paper are calculated only in the case when group are considered and allow selection of genes for normalization that are regulated, but have opposite regulation that leads to cancellation, i.e., normalization to the average of one upregulated and one downregulated genes. Theoretically, this is a valid approach. In practice its a very risky approach that may introduce severe bias. MultiD has therefore decided not to report the stability values to avoid recommending this choice (instead we recommend eliminating all regulated genes and chose the most stable from those that remain). However, when analyzing with GenEx taking groups into account the best combination of two genes is reported in the control window. This is in fact the combination that produces best stability according to the stability values. But, as said, its not promoted, because we don´t trust the cancellation of regulation on real samples.
Mikael Kubista
Posts: 455
Joined: Tue Jul 01, 2008 12:28 pm

Return to Reference genes

Who is online

Users browsing this forum: No registered users and 1 guest


MultiD Analyses

Home of the GenEx analysis software