Hello Sinem,

the presentation of results comes out directly from the method applied for data analysis and data analysis depends on the hypothesis you are testing with your experiment.

You compare 4 mice grown under the same conditions. I understand that you are interested in the variation of transcript level of candidate genes among these animals. So, your null hypothesis could be that there is no variation among the mice and that the differences observed are due to chance or error (i.e. non significant). You can immagine that your mice are randomly sorted from a homogenous populations and that the deviations are only to due chance or non controlled experimental error. Under such a model, you can compare each individual value to the population mean. This corresponds to Mario Ezquerra's suggestion.

Then you can calculate the relative expression of each mouse versus the population mean. Of course normalization by the reference genes is always necessary. If PCR efficiency is the same for all the genes you can easily apply the Delta-Delta Ct method.

At the end you can present your relative expression data as fold change respect to the population mean. Log ratio is a good transformation of relative expression data to graphically represent both up-regulation and down-regulation. Simple statistics to describe the distribution of your data are standard deviation and coefficient of variation, by which you can compare the variation of your result with that of other experiments.

However, without biological replications you would not be able to apply any statistical test to assess whether the differences are significant or not. What biological replication means strongly depends on your biological material. For example I work on poplar and I consider individual plants of the same clone as biological replications. Maybe in your experiment biological replications could be represented by separate samples collected in independent experiment on the same mice. Colleagues who work on mice could be more helpful than me about the experimental design more appropriate for you.

Hi Sinem! The data presentation depends on the aim of your measurements. Do you use any reference gene? Do you want to check if one of your mice differ in gene expression level from the other mice? If yes, you could calculate the relative expression level of the genes in individual mouse. You can use the Pfafll method for that: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC55695/

The results can be expressed in AU (Arbitrary Units).

Hi Maciej, thanks for your answer. The paper you attached is the deltadelta CT method right? That requires that i compare a condition to steady state to get "fold change" gene expression relative to steady state. What i have is 4 samples from untreated wildtype mice (steady state) and i do not compare this to anything. I have a reference gene (the housekeeping gene - in this case Rplp0). So with these, even if i can calculate the delta Ct (target gene Ct - reference gene Ct), i cannot calculate a deltadelta Ct (deltaCt condition - deltaCt steady state). I was wondering if anyone has any idea how to represent gene expression at steady state.

You can use the software LinRegPCR which is freely available in the internet. This software will give you a measure which is called N0 ("N zero"). This is a measure for the starting concentration per sample, expressed in arbitrary fluorescence units (as in the Pfaffl method). For this you don't need a calibrator sample to compare to. If you have a housekeeping gene, you can simply divide the N0 of your target gene by the N0 of the reference gene to get the relative expression value.

Dear Vivian, thanks a lot. I will give it a try!

Hoi Sinem :) Why go small when you can go big? Next time come to FGCZ and get a consultation chat on RNAseq... gets cheaper and cheaper now. And you can publish it in much more fancy way! Cheers!

:))) hey Michal, i thought why should i bother you guys with something so small.. :))

Hi, there are different softwares to do that for you i.e. qBaseplus. However for this simple model you can do it in excell as well. Chech publication of Hellemans et al., 2009 otherwise you can contact me directly.

Hi! I only need some concepts clarification. Normally we refer to steady state as the time interval where you don’t see any further variation in the expression of a gene. It would correspond to its maximum expression under inducing conditions. You can see it as the interval of time where the rate of accumulation of your gene RNA is equal to the rate of its degradation. If this is the case, how do you know you are operating in steady state conditions if you didn’t run a time course expression analysis for which you need a reference gene or a reference condition?

Hmmm, then maybe correct wording could be "untreated". There is no time course in this experiment. Single timepoint, 4 adult mice that are littermates. I am just looking at the gene expression level in the macrophages isolated from these mice. And maybe it is better if i give numbers: for instance my housekeeping (reference) gene has ct values around 20, whereas one of the genes that i am interested in has an average Ct value of 27. Now, I am aware that (considering both PCRs had 100% efficiency), my reference gene is 2^7 fold higher expressed. The question is, HOW you would prefer showing this data. I have seen some people dividing target Ct with reference Ct, some just showing normalized Ct. I just wondered if there are other ways of presenting this data.

I think what it is important is what your data is telling to the reader (most importantly to the referee). If you have an increase of X fold, then the way of representing it is a matter or aesthetic taste. Making it fancier wouldn’t add to its real value.

Keep in mind those values are only arbitrary units. It doesn’t matter if they are Cts, number of pixels if it is an image, etc…. What matter is the fold change in your gene expression between treated and untreated cells relative to a reference gene (s) whose expression doesn’t change. The most important thing is that the fold change has to be statistically significant. You need to show the SD value for that.

I have used two to the power minus dCt for my analysis on some occasions. However as the values are all relative it will be easier for others to interpret the result if you present the lowest RQ level mouse as 1 using 2 power minus ddCt.

Just a suggestion: You have 4 untreated samples and you do not have any references to compare between them? so just "create" another sample with the Ct values being the geometric mean of each gene in the 4 samples, and use this as the reference

Another option would be to set the highest of the four values to 100% and then compare the remaining three to that highest value, respectively.

In the method that I mentioned previously, you will compare to the min, max, or mean of expression levels. You wont need any control to compare. However the data should be normalized.

Why not do the ratios between the units the results gave you? I mean express the concentration you obtained from each of your sample for each gene to the concentration you got from the reference gene?

Hello Sinem,

the presentation of results comes out directly from the method applied for data analysis and data analysis depends on the hypothesis you are testing with your experiment.

You compare 4 mice grown under the same conditions. I understand that you are interested in the variation of transcript level of candidate genes among these animals. So, your null hypothesis could be that there is no variation among the mice and that the differences observed are due to chance or error (i.e. non significant). You can immagine that your mice are randomly sorted from a homogenous populations and that the deviations are only to due chance or non controlled experimental error. Under such a model, you can compare each individual value to the population mean. This corresponds to Mario Ezquerra's suggestion.

Then you can calculate the relative expression of each mouse versus the population mean. Of course normalization by the reference genes is always necessary. If PCR efficiency is the same for all the genes you can easily apply the Delta-Delta Ct method.

At the end you can present your relative expression data as fold change respect to the population mean. Log ratio is a good transformation of relative expression data to graphically represent both up-regulation and down-regulation. Simple statistics to describe the distribution of your data are standard deviation and coefficient of variation, by which you can compare the variation of your result with that of other experiments.

However, without biological replications you would not be able to apply any statistical test to assess whether the differences are significant or not. What biological replication means strongly depends on your biological material. For example I work on poplar and I consider individual plants of the same clone as biological replications. Maybe in your experiment biological replications could be represented by separate samples collected in independent experiment on the same mice. Colleagues who work on mice could be more helpful than me about the experimental design more appropriate for you.

Hi Sinem. I only need to apologize to you for misunderstanding your experimental setting. I think because you are testing constitutively active genes you can talk about their steady state RNA levels. You don't need a rat control for that. By the way, is the expression of those genes restricted to macrophages? I think you have very nice suggestions above on how to convert your CT values into relative fold of expression by normalizing against a reference gene. The reason is that if somebody obtains expression analysis by other means (Northern, semi quantitative PCR, microaaray,etc…) of those genes you are testing, he should be able to make a comparison between his results and yours. No all labs have access to qPCR machines, radioactivity, etc....

Hi Sinem, Based on your data, I best outcome I can expect may be two points. Firstly shows whether any mouse was expressed differently from the others. Secondly shows whether the expression of a gene was significantly different from another gene. The first one can be answered based on Wilcoxon paired-rank test; i.e. pair the gene expression value of a picked mouse with other three mice respectively, 12 pairs will be obtained then conduct a paired rank test. For second point, any two genes can be tested by a two-sample t test. However for this test, you should use log expression values rather than expression values, because the expression values of genes are log normally distributed. In the analysis of your data, you should covert you CT value into expression equivalent values, the simplest way is let x = 1/CT-value, hence x is equivalent to the log expression values (subject to a constant shift).