Representing qRT-PCR Data? - (Jan/30/2015 )

Hey guys, I've done my experiments several times and was wondering how to represent my data.

 

So far I've performed the same experiment about 3 times atleast, I've calculated the fold differences for each one in which two out of the three showed what I was hoping for. But how would I go about showing the data?

 

I've got the Ct values, fold differences are worked out etc. I'm looking for downregulation against a reference gene. My main idea was to represent as a bar chart of the Ct but it doesnt make sense - how would I represent the mRNA expression from the Ct values/fold difference if there is downregulation?

I've got my Ct values and here's a new question if anyone can help:

 

If I have my Ct values and 2-Delta Delta Ct, how can I represent the comparison to the controls?

Bar chart of a Ct is definitelly a bad idea for anything.

 

I suppose you are talking about relative quantification? There you usually have a "control" (calibrator) sample, and some treted/tested samples to compare with, which are normalized to reference gene, that should stay the same on both.

 

For calculation of relative quantity you can use Delta-delta Ct method or some else. Then you got a value of 1 for each calibrator sample, and values above 1 for upregulation, values below 1 as downregulation. These can be transformed into percentage values by *100 or by fold increase/decrease. You may make a chart of it (depending what other things you want in the cart, number of samples etc.) or just mention it in the text.

 

For showing 3 replicates you need to make a mean of those valuse, and standard deviation or standard error, these can be presented as error values in the graph, and from those you actually calculate if the results are statistically significant.

Bar chart of a Ct is definitelly a bad idea for anything.

 

I suppose you are talking about relative quantification? There you usually have a "control" (calibrator) sample, and some treted/tested samples to compare with, which are normalized to reference gene, that should stay the same on both.

 

For calculation of relative quantity you can use Delta-delta Ct method or some else. Then you got a value of 1 for each calibrator sample, and values above 1 for upregulation, values below 1 as downregulation. These can be transformed into percentage values by *100 or by fold increase/decrease. You may make a chart of it (depending what other things you want in the cart, number of samples etc.) or just mention it in the text.

 

For showing 3 replicates you need to make a mean of those valuse, and standard deviation or standard error, these can be presented as error values in the graph, and from those you actually calculate if the results are statistically significant.

Yes I have calculated the 2-^delta delta ct values and got my answers.

 

For one example, my claudin-3 was downregulated showing 2^-delta delta ct value of 0.112 - if I was to represent that as a percentage, would it be 17% of what? It can't be gene expression. I read up on an equation for knockdown expression but it did not make sense as I've got two beta actin samples - one control and one transfected whereas in the example it didn't follow through.

 

 

I was thinking to do relative gene expression on the y axis and represent the 2^-delta delta ct values (so 0.112) on the y-axis but I also wanted to see if it was possible to get beta actin (reference gene) on to my graph as well as my Claudin-1 control, I don't think it is possible? Otherwise if I just did this way, I would get 3 bars with Claudin-1 and my other two target genes representing if they've been upregulated or downregulated.

I have no idea what are you talking about. In relative quantification experiment, the reference gene, in your case "beta-actin" presumably is already contained in the calculation or relative quantity. You can't make a graph of it and actin again.

 

I have strong feeling, that you have bad understanding relative quantification experiment design in general. You seem not no even know what your results mean.

 

 

In a model experiment say I would use cells that are "control" or in RQ called "calibrator" and then one transfected sample.

The whole idea of relative quatification is comparison. You compare the level of gene(s) between the calibrator and your sample(s).

 

That's the first. But, as it's possible to have actual different amount of those in reaction, that would screw the comparison, another gene "reference" of "housekeeping" is selected, that should be not changed inbetween control and samples (this is very important part, that most people neglect, they just select random gene, others are using, without thinking if it is really a stable reference, because many so caller housekeeping genes may change expression after various treatments, or in cancer or so). But let's say that beta-actin is actually fine for your purpose.

So you have a calibrator and sample, and two genes you run, your target gene (claudin 3) and reference gene beta actin. You measure Cts of both, and then you use some of the common rel-quant equations.

If you don't care about efficiency that may be different between target and reference gene assays, which you don't, you use simplier delat-delta Ct approach.

In that you calculate all four Cts, Claudin (target) for calibrator and sample, and beta-actin (reference) for calibrator and sample.

You put it in the equation to get the delta-delat first:

ΔΔCt = (Ct target –  Ct reference)calibrator –  (Ct target –  Ct reference)sample

And then you make the power of two, because in delta-delta you expect both efficiencies to be equal to "2":

 

2ΔΔCT =  relative normalized ratio (or fold change)

In this ratio you already have calculated the difference between calibrator and sample, and you normalised it to the reference gene. For calibrator you always get ratio = 1 , because it is the 100%, the point of reference.
If you really got 0.112 for your sample, it means it has only 11,2 % of claudin compared to calibrator, so it's pretty downregulated.


But this is an absolute basics. You can't run it all through and don't know what are you calculationg and why, then you get some numbers and you can't say if they are right or not. Problems about type of graph are just minor.

I've got the normalised ratio for each target gene using the 2ΔΔCT equation...I have done all that work and as far as I understand <1 = downregulation, >1 = upregulation, that's fine. My original question was how to represent it in a graph, would it be fine to represent the values I got for my 2ΔΔCT calculations? (So for Claudin-1, 0.112) on a graph?

 

I dont know why you've explained the equation again, I understood it perfectly and demonstrated I did by stating I got values for them. That's not the issue, it's just presenting the data, in this case I think I would represent the values I got for relative gene expression for each target gene so will have three bars on the chart?

You were thinking about putting beta-actin into graph. Beta actin is a reference gene, that is like incorporated in every equation you make. There is no value for beta actin you could put in any graph after relative quantification. You say you can't put 0.112 in a graph, why not?
 

So clearly if you ask this, you don't have perfect understanding. But I can't know whan you do not understand unless I find why do you think that was in any case a good idea. And you ask a percentage of what is your result.. the percentage of expression of your calibrator, of course.

You ask "how should I present my data" when we have no idea what your data is, how many genes, how many samples, how many replicates. You just keep asking on what you want, but we don't know your experiment, that question can't be answered without it.
And with some obvious flaws in it, it's really dififcult to say, what you want.

So, how many target genes you have? How many samples (apart from calibrator). Do you want to have all the genes in a graph or separately for every gene..
Graphs are used to visualize better the problem, they don't tell more than numbers do, but if you make a bar graph, where there is huge difference, the picture in mind is "whoa, that's a lot". Representation is to emphasize the most important part of your experiment. 
What is it?

Ok so my experiment was to transfect siRNA into claudin-3 and monitor the effects of gene expression on Claudin-3, Claudin-4 and Twist. I definitley do understand how you cannot represent Beta Actin but thought there may be a method somehow that I may not know, I was looking for a possibility if anyone knew but it seems you can't anyway.

 

Downregulation worked for CL3 & TWIST. Cl4 was upregulated.

My 2ΔΔCT values were:

Claudin-3: 0.112

Claudin-4: 1.05

Twist: 0.20

 

Beta Actin Control Samples Ct Mean: 17.88

Beta Actin Transfected Samples: 14.78

 

I want all genes in one graph. I have an idea of how to represent it, it'd be the 0.112, 1.05 and 0.20 on a bar graph with Relative Gene Expression on the y-axis and the target genes below it. Im wondering if this is suitable presentation?

 

 

There was also something else I was looking at - the percentage knockdown - I got this from Dharmacon PDF file and which applied the equation: (1-ΔΔCq)*100. Using the information you provided before, you stated 0.112 means 11.2% is present in the transfected sample compared to the calibrator so does this mean (100-11.2) = 88.8% knockdown?

 

EDIT : I've just used this method on my values and it works - I tested it using the Dharma protocol too and it works perfectly so I know how to calculate the percentage knockdown confidently. Thank you.

This is my attempt at standard deviation of the Delta Delta Ct expression, can someone kindly check if it is correct. I followed the Dharma protocol

 

What's interesting is following this step-step guide gives me a completely different value for my calculated Delta Delta Ct value....

 

http://dharmacon.gelifesciences.com/uploadedFiles/Resources/delta-cq-solaris-technote.pdf

 

 

My Delta Delta Value is 2.59 giving a gene expression of 0.166....this is giving soemthing completely different but I cannot understand why or where I would've gone wrong.

 

Any ideas on how to calculate standard deviation from the image I put up with my data?

If anyone can help with calculating the standard deviations or seeing where I'm going wrong, please help! I've been trying to figure this out for hours :/