A lot of differentially expressed genes

Question

[gmx]

Hi

There are 8143 Differentially expressed genes (Adj. p-value <0.01) in GEO data set(GSE26459) in a micro-array analysis(by GEO2R,12 case samples vs 12 control). I want to analysis both up and down regulation with fold change. criteria for fold change value for up and down regulation ?

Thanks

[GouthamAtla]

Why don't you spend sometime in asking a good question ? "criteria for fold change value for up and down regulation ?" and the title is not related to the actual question as well.

though old, but few threads:

What Is A Good Threshold For Log2 Fold Change?

http://seqanswers.com/forums/showthread.php?t=21533

[russhh]

there isn't 12 samples and 12 controls in that experiment. There is a considerable amount of nested stucture that you have ignored. Perhaps read the paper before you analyse any dataset, read teh GEO methods/metadata, build a more appropriate statistical model etc etc. Then ask a question when you're really struggling

[Benn]

This seems more like an announcement than a question. Or am I missing something?

Why tag this 'question' with rna-seq?

Answer 1

[ablanchetcohen]

Thresholds are arbitrary and have no real biological or statistical significance. The higher the threshold, the more confidence you can have in your results, but the more likely you are to discard genes that are up or down regulated. The convention is to take an adjusted p-value of 0.01 or 0.05, and a fold-change of 1.5 or 2. But, again, these are just conventions, that have no solid statistical or biological rationale.