# Statistical evaluations for qRT-PCR - The comp. delta-delta CT method (Mar/31/2006 )

Hi all

Have anyone come across good articles or expert knowhow about statistical evaluations for qRT-PCR that they can recommemd me. I´m using the comp.delta-delta CT method and have seen that some uses a standard t-test, but I´m I bit confused though....it seems that they employ this on the CT values. Does this sound correct?

Livak and Schmittgen, 2001 (cited by ABI) suggest that statistical presentation using the raw CT values should be avoided and only aplied when looking at sample-to-sample variation.

Please let me know about your experiences on this topic

Thanks a lot

Mema

Mema,

Have you looked at the Guide to Performing Relative Quantitation of Gene Expression using Real-Time qPCR by ABI?

Pages 47-50 show how to calculate the standard deviation of dCt, ddCt, and how to get an error range for the 2^ddCt value. This allows you to present the fold change bar graph with error bars. FYI, this is basically what ABI's RQ s/w uses to analyze data using the ddCt method, except it uses std err instead of std dev.

Have you looked at the Guide to Performing Relative Quantitation of Gene Expression using Real-Time qPCR by ABI?

Pages 47-50 show how to calculate the standard deviation of dCt, ddCt, and how to get an error range for the 2^ddCt value. This allows you to present the fold change bar graph with error bars. FYI, this is basically what ABI's RQ s/w uses to analyze data using the ddCt method, except it uses std err instead of std dev.

Thanks for your reply soluene

I have looked on the ABI guide. As I remember they give an example using liver, bladder and kidney with liver as the calibartor. For the fun of it (cannot recall the excact values/results of the example) lets say that they measure a relatively gene expression for any given gene of 10 in kidney compared to liver (calibartor, rel. epxr. =1). This of course means that the gene is transcribed 10 times higher in kidney compared to liver, but how do you verify this statistically wihtout doing plate replica. Do you compare the raw ct values of kidney to liver? using a t-test?

What if the relatively gene expression was less obvious e.g. 1.78 in kidney compared to liver/calibrator and you need to verify that the gene is actually significantly higher expressed between tissues?

Just can´t figure it out...mayby I haven´t seen the light yet!

mema

Mema-

I'm sorry, I don't have an answer for you. We just show the error bars in our reports, but I don't have to determine statistical significance. Sorry my answer was off from your question! I hope someone else can help you.

I would checkout Wong and Medrano, 2005, Biotechniques...they have a review titled Real-Time PCR quantitation or something to that effect. Biotechniques is a free publication that you can access online if you register.

They reference several papers that provide free analysis software. I personally use qgene.

I would use a gene like beta actin or rRNA for relative quantitation.

-Matt

I recommend this reference:

Yuan JS, Reed A, Chen F, et al. Statistical analysis of real-time PCR data. BMC Bioinformatics. 2006;7:85.

They defend the idea of comparing statistically at the DeltaCt level, before exponential transformation. But the final statistic test should be depending on your experimental model (t-test, multiple ANOVA, Regression...)