You could do absolute quantification, which involves making a standard curve of known concentrations of your product for each of your targets. You then measure your samples relative to the standard curve in the same run. Your results can be normalized by taking into account product size (99 vs 150 base pairs) and by normalizing to a housekeeping gene that is measured in a similar fashion. I hope this helps.

I can suggest lineReg, LRE, Cy0... It depends on the type of PCR inhibition

Read papers by Vandesompele et al, also described here http://medgen.ugent.be/~jvdesomp/genorm/. Instead of using a delta delta Cq (where the implied efficiency of both primer pairs is perfect) you run a standard curve to determine primer efficiency and can thus use two disparate efficiencies to calculate differences in expression. I believe they also have free software available for academic use to assist in this, and newer software for BioRad qPCR instruments already has it loaded (if you're lucky enough to be running on a BioRad platform).

This is a good method of choice in that case:

Q-Gene: processing quantitative real-time RT–PCR data by Perikles Simon

With LinRegPCR (freely available) you can choose to analyse your samples with their individual efficiency. However, when you do that, you will have very high variation (also shown in the original article by Ramakers et al. 2003 and Ruijter et al. 2009). Maybe you can also find information about other programmes in the following article that compares the most common freely available qPCR analysis software (Ruijter et al. 2013):

http://www.sciencedirect.com/science/article/pii/S1046202312002290

If the samples show different efficiency for one gene than for another (e.g. primers for one gene are less effective than for another), then I suggest the relative standard curve method. It requires running a standard curve on each plate but it is worth it as it helps to get around the fact the efficiencies are different (better than designing and testing new primers).

However, if the samples show different efficiencies for the same gene, then it is likely that the RNA purification or subsequent RT reaction (or both) were not the same for all samples. E.g. if some of the samples had some residual phenol in them, the RT reaction would not be so efficient and that could also affect the qPCR efficiency. In that case the cDNA samples are useless, as one doesn't know what part of the expression reflects "real" expression and what part is an artifact of the procedure, and even more, the proportion of the artifact will be different for different samples. In that case no matter which method of analysis is chosen, it will not give you real results.

I think the equation you are looking for is written out in Pfaffl's 2001 paper, it's equation #3. It allows a calculation with different efficiencies of calibrator, sample, target and reference genes. Open access: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC55695/ If you want to determine efficiency on a sample-by-sample basis, you can fit the amplification curves to a (preferably) 5-parameter log curve, and determine the inflection point. This is the Cy0 method (I think). I think you must always run the reactions to a plateau in order for a log-fit to be accurate. Lots of info about Cy0: http://www.cy0method.org/

REST will analyse your samples with different efficiencies of amplification and is free for academic use, as mentioned by Leta .

http://www.gene-quantification.de/download.html

As Eric Patterson said before, I suggest Pfaffl method too. In this method, you can normalize each sample independently. All data that you need are similar to delta delta Cp method but its equation and therefore normalization is different. Also pfaffl method give an accurate result than delta delta Cp method.

I am planning to use the Pfaffl equations as well; I have attached a copy of the 2001 paper for ease of access. Good luck with your work.

I think Eric Patterson is closest to answering the initial question with the Cy0 method by Guescini and Sisti, et al. The other 'variable' considered in the Cy0 method is the 'inflection point' as determined by 5-parameter Richards-equation modeling.

The Cy0 calculates the 'best' inflection point of each log-linear region of each amplification plot (when y-axis is in log of relative fluorescence units [RFU] and the x-axis is in Cq units) and draws a tangent to that calculated 'best' inflection point. This tangent is extended to intersect at or near the x-axis (where y = 0, thus "Cy0"). This way the 'pitch' of each each amplification's log-linear region is 'precisely' taken into account. And for multiple samples, where the tangents intersect [at or near] the x-axis gives you the 'true' differences between the Cq (now termed "Cy0") values of the samples in relation to one another.

Though not immediately obvious, the Cy0 method is still taking efficiency into account - since a 90-degree rotation (clockwise) of any qPCR/RT-qPCR amplification sigmoidal plot (of log of RFU vs Cq) is a standard curve (for every single reaction) anyway. The 90-degree clockwise rotation gives you Cq on the y-axis and log of RFU on the x-axis -- which is directly analogous to serial-dilution-based standard curves plotted with Cq values on the y-axis and log of relative sample dilution (or log of copies if you're lucky enough) on the x-axis.

When the slope of the Cy0 tangent drawn through the log-linear region of an amplification curve = -3.3219, it indicates a 100%-efficient amplification reaction; the same slope value that would be obtained from a 100%-efficient serial-dilution-based standard curve.

The Cy0 method merely attempts to be certain that the inflection point of each amplification's log-linear region has been determined correctly (or 'precisely') so that the slope of the tangent line drawn is truly indicative of each amplification reaction's log-linear regions's slope (therefore efficiency) as it is interpolated to the x-axis to apprehend "Cy0."

As the familiar equations E = [10^(-1/m)]-1 and Eamp = 10^(-1/m) hold for serial-dilution-based standard curves, the Cy0 method holds the same relationship (between slope and efficiency) true for each amplification curve's log-linear region as well.

A recent paper showing the Cy0 method to be among the top contenders (if not the top contender) for accurate qPCR/RT-qPCR quantification in many circumstances, is the following paper in "Methods" from late 2012:

Evaluation of qPCR curve analysis methods for reliable biomarker discovery: Bias, resolution, precision, and implications

Jan M. Ruijter, Michael W. Pfaffl, Sheng Zhao, Andrej N. Spiess, Gregory Boggy, Jochen Blom, Robert G. Rutledge, Davide Sisti, Antoon Lievens, Katleen De Preter, Stefaan Derveaux, Jan Hellemans, Jo Vandesompele

I don't know a single analysis software that deals correctly with uncertainties in the estimated efficiencies. Therefore I would NOT recommend such software. However, when the uncertainties are considered, you will see that the precision of the estimates is usually so bad that the whole result is pointless. The only solution remaining is to ensure (and/or reasonably assume) that the efficiency in fact is optimal (at least in the early cycles c100% are estimated, sometimes as 110% or so. Nice to know that these excessive 10% must be some technical problem, but what if you mesaure, say, 80%? is there also a bias?(*) What is the "real" value? It can well be anything between 70% and 90% or even worse. Plugging 80% into the formulas to calculate fold-changes can thus be drastically biased (errors in the efficiency estimates exponentially effect the results!).

(*) I talk about "bias" here because in efficiency determinations there often is a large error caused by the experimental setup (buffer conditions, instrument, fluorescence chemistry, algorithm of background correction, ...) what can be seen as a "systematic error" (hence, bias). The residual, stochastic or random error is much smaller (**), and -if at all!- an error is taken into account, then it is this stochastic error, because the bias is hard (or impossible?) to estimate.

(**) I have seen estimates for the efficiency of 110% plusminus 2%. In suc a case, there is an obvious bias of around +10 percent points, but the stochastic error is as little as 2 percent points (depending on the setup it may be even smaller).

Hi everyone,

I am new to Real-time PCR. I had done RTPCR for my samples using Sybergreen (File attached). Can you guys please tell me whether my Ct values are ok or not?

Thanks in advance

Quite nice. If differences between repetitions don't excede 0.5 cycle, it is excellent.

Thank you Maciej