I compare control with treated sample and use single HKG
First I was using ddCt method, but then decided to switch to more precice Pfaffl method to include my efficiencies which are not the same. However, they are also not the same within same gene in control and treated.
GOI (control) = 0.93 GOI (treated) = 0.88
HKG (control) = 0.97 HKG (treated) = 0.91
Which one can I use for calculations (in REST)? I was thinking about average or smallest?
Or maybe I should still use ddCt method as my validation works? (standard curve of ct(GOI)-ct(HKG) produces trend line with slope
You can use all for caluclations, just in Excell you have to substitute 2 to the power of ddCt
GOI control : 1.93 to the power of ddCt
GOI treated : 1.88 to the power of ddCt
etc.
You can use all for caluclations, just in Excell you have to substitute 2 to the power of ddCt
GOI control : 1.93 to the power of ddCt
GOI treated : 1.88 to the power of ddCt
etc.
I would suggest you try a few other HKG to try to identify the one that would be the most constant throughout your groups. You can check here for inspiration:
http://www.sabiosciences.com/rt_pcr_product/HTML/PAMM-000A.html
How are the confidence intervals for the efficiency estimations?
How are the efficiencies determined? Are they independent from the ct values?
Why I ask this: The determination based on the slope of the log-linear segments of individual curves can be quite misleading, because of problems with an appropriate background/trend correction. This is all quite heuristic and arbitrary. A dependency between ct and efficiency (calculated based on these slopes) is indicative for a problem in background/trend correction.
The problem with this method is that it uses signals that are already close to the plateu phase and they effectively ignore the process in the early cycles before, although just those cycles impact the ct values.
Side note: Pfaffls method is not more precise. It is rather less precise (you must consider the uncertainty in the estimation of the efficiencies!). It may be more accurate, if the efficiencies are dissimilar.
I run external standard curves for all my house keeping genes and genes of interest for every PCR run with each primer/probe set.The standard curve are generated from reverse transcribed whole organ RNA that has been aliquoted and stored at -80. Rather than use ddCT, this allows numerical values for each primer/probe set to be determined from the standard curve (for GOI and housekeeping genes) and takes into account the efficiency of each primer/probe set for every PCR run for each experiment. I would also advise using at least three housekeeping/reference genes and taking the geometric means of the data achieved relative to each housekeeping gene. Although this approach is a little more costly and time consuming, its gives very tight reproducible data and allows a comparison of data across many experiments over a long time period if the RNA for the standard curve is properly aliquoted and stored
Thank you all for your answers.
I believe my problem in different efficiencies between treated samples and untreated lies in RNA purity. I always use RNA RIN values above 8, but I don't precisely control for salt/organic contamination. I perform magnetic beads purification and take from there..
Isabelle R Miousse , In this case using few HKG wouldn't help (I actually use 2, but in all cases in treated samples efficiency is less)
Jochen Wilhelm, I estimate efficiency based on standard curves (automatically in BioRad software). why can't I trust calculated values if it shows good correlation with different concentration? what method do you suggest?
Sean Wyatt, I was thinking of standard curve analysis, but it seems that my control (untreated) is my external control. I work with bacteria, and wild type is my "untreated" and one of the gene-deletion mutant is my "treated". So in the case what would be my another control?
So I always run standard curve for HKG, GOI for both control and treated and from there take mid-point, let's say 1 ng from 100-0.001 dilutions for calculations.
Jivko V Stoyanov, do you mean we substitute all 2 to corresponding efficiency? That sounds reasonable...
Hi Irina,
The bottom line is you should not be getting different efficiencies for different samples. If you are, like you said you probably have some sample contamination problem and you should address this first and foremost. If your sample is contaminated with phenol or salts it will affect PCR efficiencies in quite unpredictable ways and your data will not be reliable. Phenol contamination is usually easy to spot when you use nanodrop or another full spectrum spectrophotometer for your RNA concentration measurements. That being said, your efficiencies are not hugely different, so like Jochen said it could be just a question of imprecise estimations of your efficiencies. Have you repeated your standard curve runs multiple times and always gotten the same values for efficiencies? How many samples did you use to make a standard curve - was it a single sample for control and single sample for treated? Or did you pool your samples for each treatment group and generate standard dilutions from these? Anyway, hopefully you can get the issues resolved.
Best of luck,
Pawel
please note that the GOI/HKG ratio is kept constant (equal, actually) between control and treated... this is a good guarantee to keep adopting the ddCt method without troubles.
- Badges
- Science topic