Ct variation/Ct limit/rē values/efficiency errors - (Feb/26/2010 )
Hey guys, I have some important questions in order to accurately evaluate some real-time data from students in my lab. Here they go:
If analyzing real-time pcr data, I was wondering about these important points:
1) Suppose you have triplicates for each sample/standard: the Ct-values of these three replicates are never exactly the same, but which degree of variation is allowed and should you remove wells of which the Ct-value is very different from the other 2? Is the cut-off around Ct +/- 0.15? And are data with Ct +/- 1.0 reliable or not at all?
2) If you read data or have wells which have Ct-values of more than 35, are these data or wells usable or is 35 the maximum allowed Ct-value? (I thought more than 35 was comparable to the baseline threshold)
3)When real-time data have correlation coefficients (rē) of 0.91-0.95, are these data reliable? (I thought data should have at least 0.98 rē)
4) If you have data with efficiencies of more than 100 % (standard curve slope > -3.33, for example -2.90), are these data reliable? Is the > 100 % efficiency due to improper settings of the threshold/baseline or due to pipetting mistakes?
If you have answers to any of these questions, that would be extremely helpful!
Thank you in advance and greetings!!!
Wozzels
wozzels on Feb 26 2010, 07:15 PM said:
If analyzing real-time pcr data, I was wondering about these important points:
1) Suppose you have triplicates for each sample/standard: the Ct-values of these three replicates are never exactly the same, but which degree of variation is allowed and should you remove wells of which the Ct-value is very different from the other 2? Is the cut-off around Ct +/- 0.15? And are data with Ct +/- 1.0 reliable or not at all?
2) If you read data or have wells which have Ct-values of more than 35, are these data or wells usable or is 35 the maximum allowed Ct-value? (I thought more than 35 was comparable to the baseline threshold)
3)When real-time data have correlation coefficients (rē) of 0.91-0.95, are these data reliable? (I thought data should have at least 0.98 rē)
4) If you have data with efficiencies of more than 100 % (standard curve slope > -3.33, for example -2.90), are these data reliable? Is the > 100 % efficiency due to improper settings of the threshold/baseline or due to pipetting mistakes?
If you have answers to any of these questions, that would be extremely helpful!
Thank you in advance and greetings!!!
Wozzels
Hi Wozzels,
I might be able to answer ur first 2 questions.....
1) triplicates shpuld ideally be the almost same........with Taqman, a variation of +/- 0.5 is considered goos, whereas for Sybr green...+/- 0.75. 1.0 is too high..........if its a triplicate set frm the same naster mix, then, of course, u should suspect pipetting errors, or primer dimer formation.....
2) n i think u shud discard the data with Ct values higher than 35...
3) about r value, i would myself like to have an answer...
Here are my two cents:
1. You should never remove replicates unless it is clear that they are outliers. A Ct difference of 0.15 is not an outlier, while a difference of >1.0 could be an outlier depending on how tights your other readings are.
2. It is perfectly fine to use a Ct value of 35 if you run 45 to 50 PCR cycles. The only reason why you would not use a Ct value of 35 is if you only run 40 PCR cycles. This is because you cannot really tell if the amplification is real when you only have 5 PCR cycles beyond the point where the Ct is calculated.
3. Any qPCR run with an r2 lower than 0.98 is for all intended purposes useless (not reproducible/tight enough). In my experience I've never used any qPCR assay that gave me an r2 lower than 0.99, and I have always strived for an r2 higher than 0.995. In short, an r2 higher than 0.98 is ok, although sloppy, an r2 of 0.99 is better, and an r2 >0.995 is good technique.
4. Data with efficiencies higher than 100% are rare, but possible, and they are likely not due to threshold/baseline settings (although if done incorrectly, settings the threshold/baseline manually can give you all sorts of weird results). There are many ways you can get efficiencies higher than 100%, but the most common is the presence of an inhibitor in your sample.
My recommendation: test different qPCR assays, for the same gene, to find the one that gives you the best results (highest r2, closest to 100% efficiency, etc), and only then use it to analyze your samples.
