Which Parameter Should Be Used For Comparativemarkerselection In The Genepattern Software?

Question

[jlshi.nudt]

Hi, I am using Gene Pattern to find differential expression genes. The ComparativeMarkerSelection module provides a list of statistic parameters such as rank,Score, Feature P,Feature P Low/High, FDR(BH), Q Value, Bonferroni, maxI, FWER and fold change. I just know these parameters are similar with p value, but I don't know the exact meaning of each one.

My question is how to select differential genes? which parameter should be used? Thank you.

[munch]

I'm not familiar with GenePattern, so I dont know how the Rank or Score is being calculated. A possible threshold can be the absolute fold change (What is a good threshold for log2 fold change?). Take also a look on the p-values after multiple testing correction (FDR), they should be smaller than 0.05 to be significant. This is also worth to read.

From the website:

A high score indicates correlation with the first phenotype (upregulated in ALL) and a low score indicates correlation with the second phenotype (upregulated in AML).

Answer 1

[BHillMeyers]

Hello,

Please read the module documentation to understand more about the parameters. You can access this via the "documentation" link in the upper right hand corner of the module page on the server.

I've also included a link here: http://www.broadinstitute.org/webservices/gpModuleRepository/download/prod/module/?file=/ComparativeMarkerSelection/broad.mit.edu:cancer.software.genepattern.module.analysis/00044/9/ComparativeMarkerSelection.pdf

The quick start listed by munch is a good introduction, but you may also find the Differential Expression Protocol useful (http://genepattern.broadinstitute.org/gp/pages/protocols/DiffExp.html)

and this indepth article: http://www.broadinstitute.org/cancer/software/genepattern/gp_guides/indeptharticles/sections/cmsForDiffExp

Please let us know if you have any further questions.

Best

-Barbara Hill Meyers Cancer Program Informatics Development The Broad Institute of MIT and Harvard