qRT-PCR: how to average multiple runs? - (Aug/19/2008 )
I am performing a relative quantification RT-PCR experiment with a gene of interest and a housekeeping gene in an experimental condition and a control condition. I ran all samples in triplicate. I have done the experiment three times. I have standard curves for both genes and I'm relating the gene of interest to the housekeeping gene. I'm then normalizing the experimental condition to the control condition. I am getting stuck with the math during the final step of trying to average the three experiments. Here's how I am proceeding.
I compare all samples to the standard curves to get an 'amount'. I then average the three replicates for each sample and get the standard deviation. I then divide the average for the gene of interest by the average of the housekeeping gene to get a relative amount. I propagate the standard deviation by multiplying the quotient by the sqrt of the sum of the squares of the CVs (st dev/mean).
I then normalize my experimental sample by my control sample and once again propagate the standard deviation.
That's all fine and dandy. The problem is that I did the experiment three times and I need to take the average of the normalized values. For instance, I have the following:
3 experiments (experimental/control): 1.1+0.1, 1.5+0.2, 0.9+0.05
How do I average these and propagate the errors? If I just ignore the errors and average then I would get 1.17+0.3, but since there is error in the amounts that I am averaging, don't I have to propagate them somehow?
in this paper they talk extensively about comparing data from different runs and error propagation. its very helpful, even if you don't have the discussed analysis software.