qPCR Calc Help - (Oct/25/2011 )
Please excuse my basic questions, but I'm quite new to qPCR and trying to learn as I go.
One of the experiments I am carrying out is to compare the expression of Gene X between two samples (disease positive vs. disease negative). I have carried out qPCR for the gene of interest, as well as an endogenous control (GAPDH in this case) and generated Ct values at an appropriate threshold. The average Ct for GAPDH in samples 1 and 2 was within 0.2 of each other.
From here, please correct me where I deviate from correct protocol, or show misunderstanding ........
<*>I first calculated the average, standard deviation and %RSD for the replicates of each reaction.
<*>Next, calculated the deltaCt of Gene X for each of samples 1 and 2, by subtracting the Ct for GAPDH from that of Gene X. This gave me a deltaCt for each of sample 1 and 2.
<*>Calculated Relative Expression for samples 1 and 2 using the equation (2^-dCt). This gives me a value representing the expression level of Gene X in each sample. These values may then be compared to each other as relative expression values between samples 1 and 2.
I get a bit muddled from here on .....
<*>I was then asked to calculate the delta.deltaCt by subtracting the deltaCt of sample 1 (disease) from that of sample 2 (non-disease).
<*>Calculate a comparative expression using the equation (2^-ddCt). This gives me a single number indicating the expression of Gene X between samples 1 and 2.
Is the logic of all this correct or have I misunderstood or blatantly got everything wrong?
Many thanks to all for input.
Im also new to qPCR and it would be really amazing if someone could answer this whole thing properly.
What about error bars? What Ive seen people doing is taking the SD value from a sample.. and SQUARE SUM it with SD value from normalizer. And then calculating 2 ^ this value.. That would be max bar and min bar from average expression from one biological replicate.
But what happens when you are calculating error bars for 3 biological samples?
Why don't you try using a excel macro ? or do you have to calculate it manually ?
Here's some free tools for qPCR calculations: http://www.gene-quantification.de/download.html
REST is a good one. It will calculate SD for you too.
Answer to your initial question is yes, its correct, you are using Livak method for calculation of ddCt, people also use Pfaffl method, where you get dCt by subtracting treated and control samples first for gene of interest and reference gene and then subtract dCt for target gene from reference gene to get ddCt. then 2^-ddCt to get the single number.
EIther way you get a single value referred to as fold change in expression relative to reference gene.
Many thanks, I can rest a bit easier now. Many thanks for the link also.
In your experience, is there advantage to one method over the other (Livak / Pfaffl), or can data derived from both be compared to each other?
Either one should give you the same fold difference. I have seen more papers use Pfaffl method, also the REST software is based on that method.
I don't really know what advantage one has over the other, but if I were to guess based on data, then I would say if you have high level of consistency in your Ct values of house keeping gene in all of the biological samples then using Pfaffl (normalization with Ct values of house keeping gene at the end works fine) seems more logical. However, some level of variability between biological samples/replicates (I mean from one experiment to another) in Ct values of house keeping gene (which should ideally not happen, but sometime it does happen, as long as it is not high variability), it might be better to normalize your target gene expression ct value with house keeping gene ct value before you normalize between sample groups (to control group) i.e. use Livak method. This way to avoid variation between experimental runs, and you get a better SEM (error bars) once you are done with your statistics. So bottom line is that using either method should only vary your error bars at the end, not the fold change in expression.
I hope this makes sense, this is something that I came up with based on my experience of qPCR data analysis, I may be wrong too.
The way I understand it, delta-delta Ct method doesn't account for different reaction efficiency of each primer pair (or it doesn't account for the efficiency at all and assume it's 100% when using 2^-ddCt). Pffafl method can do that, thus if your efficiency is not 100% in both target gene and housekeeping, Pffafl should give more accurate results. For this reason Pffafl is prefered among the leading ..guys that are really good with qPCR Otherwise the substraction wibbly-wobbly in both is equivalent.
But I heard an opinion, that you should design primers to have ~100% efficiency and then you can use ddCt without worrying.
As with the errors of final results it's a bit complicated. I'm not really fond of statistics, so I let the machine (LightCycler480) to calculate this when I can. Also is you have like 3 biological replicates then you just make error bar from those three results not taking in account the SD of each, that's what a colleague told me.
I have also been puzzled at the lack of a complete 'standard' for qPCR statistical analysis; especially as I am a relative newbie with real time PCR! I found the ABI relative quantitation guide helpful in trying to come up with standard deviation of the RQ values, but I'm unsure whether this is the best method as it does not mention the Pfaffl method.
Would it be valid to carry out a duplicate qPCR with samples in triplicate, taking the average efficiency-corrected fold change from this and carry out normal 1-way ANOVA?