Protocol Online logo
Top : New Forum Archives (2009-): : PCR, RT-PCR and Real-Time PCR

RT-PCR analysis - (Feb/24/2015 )

Hey guys I'm looking for some advice in regards to real time analysis. I analyse my data with the 2-deltadeltact method which I am confident using. My problem comes with trying to combine the data with 3 independent experiments. As I am working with the 3T3-L1 cell line the experimental variation with each independent experiment is huge. The data for each experiment is tight and therefore assay performance is good. The pattern of expression for genes investigated in each individual experiment is the same however the fold change is often very different and when combined the data looks terrible. 

 

I have used a method to eliminate experimental variation and I am looking to see if anyone else does this or whether this is an acceptable way to approach this problem. 

 

This is a simplified version of data but say in one experiment (post 2-deltadelta analysis) I have for a specific sample an MX value of 10 and Fabp4 value of 100 and in a separate independent experiment an MX value of 100 and FABP4 of 800 is it acceptable to eliminate experimental variation by utilising the inital MX data to alter the second i.e. in this case 10/100 = 0.1 X 800 = 80 so that when combined the data values for FABP4 are 100 and 80 rather than 100 and 1000.

 

I have done this for all my data and it sort out my problem I would just appreciate if anyone knows whther this is acceptable or not (hope my explanation makes sense)

 

Thank you 

-markire05-

If you get a different fold change for the same gene in each replicate, then you don't have the same pattern of expression.

 

If you mean, it seems that all genes are multiplied by similar constant in each experiment (like 10 fold increase for first gene in one replicate and 100 fold increase for another, with all the other genes following this pattern) then you seem to have a serious problem with the reference to which you are normalizing, because the relquant experiment should normalize already for this variation you are trying to get rid of. Considering of course that calibrator and samples have been independently cultivated at the same time and place, not that you.. I don't know, use the same calibrator sample for all replicates or something.

 

You can't just divide the data, that should be very similar (because they are exact replicates), that is not a correct manipulation with the data. There is something wrong with your design, or results.

-Trof-