Real-time PCR....Making sense of it all? - (May/18/2007 )
Hey everyone,
I guess my question this time is more theoretical/philosophical.... I'm relatively new to the real-time PCR field, and just find it almost overwhelming the amount of techniques/chemistries out there, especially when it comes down to "crunch time" when you're ANALYZING all of these Ct values to see what gene X is doing in the treated samaples (increases/deceases), etc. People on my own floor seem to have 50 different ways of analyzing their data, some people use Pfaffl method and express as a ratio (others express as %tages), others use delta-delta Ct, others people I've seen here use equations they seem to derive out of nowhere..... so I'm now wondering.... how cautiously must one approach this technique?? When will the "gold standards" bet set for real-time PCR to set everyone at a level playing field when it comes down to publishing the results?
Don't get me wrong, I love the real-time PCR (qRT-PCR is what I'm doing), and I like it a lot better than the semi-quant method, but it's intriguing to wonder how much data is being published out there with either 100% incorrect info, over-interpretation of effects where people want to see them, or just general "mess-ups" with using all of these equations.
Just wodering people's thoughts on this.
Thanks!
MapleCanPCR
PS: This is just my 2nd post here on the Bioforums, but want to say what a great resource this is!! I wish I found out about it earlier!
You're right. There are a lot of problems with the accuracy the method is used with. I know a lot of studies published, just using a cycler and ooops, something significant comes out of it. I think, during the last years, the reliability of the method and our understanding on how to interprete the results has increased. Still, a lot of different methods for the analysis exist. Paffl's equotation is rather simple but easy to understand. Other approaches are much more sophisticated, nevertheless, they might reflect the uncertainties of the method even better. I think it is good scientific practice to state very clear waht has been done (like - what was measured on the same plate, where standard curves (relative or absolute) used, if yes on every plate? Where the primer efficiencies stable? Articles should state efficiencies or the slope, and so on). Those paper providing all of these information have nothing to hide and are probably much more reliable than those stating solely
To make the whole story short: real-time PCR is a wonderful method. It can be misused. Show everybody that you use it right.
I'm doing qRT-PCR, too and there are many people here doing it for different purposes. The best thing is that: It can be used for many purposes. However, the problem is that we still don't know much about the real nature of the technique. Only the data from the exponential phase is used to calculate efficiency and Cp but it's just a piece of the data. The power of the technique is not utilized 100%.
So I personally don't want it to be standardized. (yes, I'm one of those using the equations). Let people think about novel approaches and publish them as long as they prove the reliability of the technique statistically. In the end, if you want to abuse something, you can do it anyway.
Well I'm relatively new to relative quantification on real-time and from the methods I looked up so far I would say, that Pfaffl and delta-delta actualy don't differ very much, when efficiency is linear and high. I mean if you have 10 fold increase with one, then you would get like 7 fold or 11 fold with other methods, but the fact is there is some like-10 fold increase is obvious.
It's something different when there are really small differences in expression, but that could also fall to the intravariability of the assay and the whole fact that we dont't know exactly what happens inside the well.
I think I will use equation in the software module for relative quantifications that gives identical results like Pfaffl (only that program does them, not me ;) and uses 2nd derivative method for determining Ct and check it for the importatnt samples with other methods. If the result ids very different, then I'd try to figure out the best fitting, otherwise it really doesn't matter.