Comparison across multiple plates - (May/01/2007 )
By using the relative standard curve method with 5 different concentrations for making the standard curve plus a set of NTC sample (non-template control) and with every sample in triplicates, one 96-well plate is only enough for 10 samples. If I have more than 10 samples to compare, it seems to me that I need to run them on multiple plates. Now I'm wondering whether I can compare the results from multiple plates directly and if so, how.
After I did some research, I think one way is to place a common calibrator on every plate. All the final results (in this case, all the input amounts normalized to the internal control) will be calibrated with the same calibrator.
But I'm also wondering that as all the results will be normalized to the internal control and the normalized values are actually ratios (which means "in sample x, the amount of gene Y is Z times of the internal control"). And if we assume that the level of the internal control is constant across all the samples, we should be able to compare these ratios from multiple plates with each other directly. In this case, it seems to me that I don't need to place a common calibrator on each plate, and I can save the space for one more sample.
Any comments?
Dadi
are you running the same standard curve on every plate?
and another possibility of saving space is multiplexing.
and another possibility of saving space is multiplexing.
I'm running the same standard curve on every plate. As we're using Sybr Green, multiplexing is not an option and the extensive optimization for multiplexing is also not trivial.
I got some answers for the same question back from ABI. They told me that usually there are two options. One is to include a common "positive control" on every plate, which by my understanding could be the "calibrator". As the calibrator is always set to be 1 at the end of the relative quantification, the results from multiple plates could be compared validly. The second one is to just rely on the fact that all the quantities at the end have already been normalized to the internal control, i.e. they are expressed in ratios, and just compare these ratios directly and not to worry about any "common positive control".
For my own case, I'd like to include a common calibrator on every plate for better accuracy.
The usual method is to select the sample with the lowest amount of your GOI, assign this sample arbitrarily as "1" ("calibrator") and display all the other samples as n-fold change to this calibrator. therefore you don't get results <1. [another possibility, if you are comparing treated and untreated groups is to assign your untreated control as calibrator]
obviously, this doesn't work if you want to compare multiple plates. so, you can select one of your samples, declare it the calibrator and run it on every plate. you may get n-fold changes <1 and you lose space for one sample on every plate. Maybe you can calibrate to your standard curve? If your standard curve contains pooled sample cDNA it should resemble a cross-section of your GOI expression profile in your samples and its already running on every plate. just one consideration.