Hi all Am curious to know why the recommended cutoff to remove non-variable genes is an IQR of 0.5. What is the significance of this number? Also, why use IQR as opposed to any other measure of variability, like SD? Cheers -Suraj Suraj Menon PhD Student Department of Pathology Henry Wellcome Building School of Medicine Cardiff University Heath Park Cardiff CF14 4XN Tel: +44 29 2074 3979 Email: MenonS1 at Cardiff.ac.uk

Suraj Menon wrote: > Hi all > > Am curious to know why the recommended cutoff to remove non-variable genes is an IQR of 0.5. > What is the significance of this number? Hi, It is the median (not the value of the IQR), so you are taking the top .5 (by variability). A bug in an early version used the number directly, but this has long since been found and fixed. > Also, why use IQR as opposed to any other measure of variability, like SD? Use what ever you want. IQR tends to be more robust to outliers, which at least at one time were somewhat prevalent in gene expression studies (they are less so now as all relevant technologies have improved). best wishes Robert > > Cheers > -Suraj > > Suraj Menon > PhD Student > Department of Pathology > Henry Wellcome Building > School of Medicine > Cardiff University > Heath Park > Cardiff CF14 4XN > > Tel: +44 29 2074 3979 > Email: MenonS1 at Cardiff.ac.uk > > _______________________________________________ > Bioconductor mailing list > Bioconductor at stat.math.ethz.ch > https://stat.ethz.ch/mailman/listinfo/bioconductor > Search the archives: http://news.gmane.org/gmane.science.biology.informatics.conductor > -- Robert Gentleman, PhD Program in Computational Biology Division of Public Health Sciences Fred Hutchinson Cancer Research Center 1100 Fairview Ave. N, M2-B876 PO Box 19024 Seattle, Washington 98109-1024 206-667-7700 rgentlem at fhcrc.org

On Wed, Apr 30, 2008 at 9:47 AM, Suraj Menon <menons1 at="" cardiff.ac.uk=""> wrote: > Hi all > > Am curious to know why the recommended cutoff to remove non- variable genes is an IQR of 0.5. > What is the significance of this number? There is not any recommendation that I know of. You will need to define your own cutoffs, generally. > Also, why use IQR as opposed to any other measure of variability, like SD? You can choose any measure that you like. IQR is less sensitive to outliers than SD, generally. Sean