The meaning of a Ct value depends on the amplified product and on the run and the instument settings. You can NOT compare Ct values of different genes (well, it surely works semi-quantitatively: when on Ct of gene A is very much larger than the Ct of gene B, this does in fact indicate that there were fewer copied of A compared to B).

If you want any useful measure from your experiment (having only tumor samples but nothing to compare to), you will have to make an absolute quantification, e.g. how many mRNA copies of your gene of interest are in a sample (possibly calculated as "per cell", or, in the absence of this possibility, as "per copy of GAPDH mRNA" - so GAPDH would have to be quantified absolutely, too.

Hi! 

The best way to compare such reults is to evaluate deltaCt an then calculate

2^-deltaCt. You will get really small numbers as results but You can multiply them all by 1000 and then You can easily compare them, do statistics and present data on graph :) 

I hope that I helped :)

Good luck!

Thank for your advice Ms. Katarzyna,

I am mere an amateur in this field. I always see that people use 2 ^delta ((delta-Ct of treated)-(delta Ct of non -treated). So how can I do. Would kindly guide me more detail for me?

Another one thing, my data doesn't have statistically significance when I use t-test SPSS.

Lower the ct value higher is the expression of gene. If the mean of target gene 's Ct value > the mean Ct of GAPDH , it means target gene expresses less than GAPDH.

The meaning of a Ct value depends on the amplified product and on the run and the instument settings. You can NOT compare Ct values of different genes (well, it surely works semi-quantitatively: when on Ct of gene A is very much larger than the Ct of gene B, this does in fact indicate that there were fewer copied of A compared to B).

If you want any useful measure from your experiment (having only tumor samples but nothing to compare to), you will have to make an absolute quantification, e.g. how many mRNA copies of your gene of interest are in a sample (possibly calculated as "per cell", or, in the absence of this possibility, as "per copy of GAPDH mRNA" - so GAPDH would have to be quantified absolutely, too.

Thank for your all advice.

@ Jochen: Can I make an absolute quantification without standard curve. At present time all data I have only Ct mean of 30 samples of reference gene and interest gene.

No, that's not possible without a standard curve.

Actually you will have to create standard curves and re-measure your 30 samples again (together with the standard curve).

If you can only re-measure some of the samples (possibly because some are used up already), then you can can adjust the ct-values of your previousely measured samples by the difference in the ct-values you observe between your previous and your actual measurements.

----

Example:

You have the samples A, B, C, ... and the previous ct-values were cA, cB, cC, ...

Now you measure a standard curve and onyl sample A. The new ct value you get for this sample is cA'. You can now correct the ct values of all the other samples (B, C,..) by adding the differencce (cA'-cA) and use these corrected ct-values for the absolute quantification.

----

Note that this is a work-around! It would be clearly better to actually re-measure all your samples, so that they are measured under exactly the same conditions as the standards.

Thank for your kindly response, Jochen.

Maybe my data has to give up because in my condition, re-measure is impossible.

:(

You could use the following paper to guide through your data analysis:

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC4280562/

@ Ms. Carolina,

So many thank for your suggest. But same as the  2-ΔΔCT method, we must have control sample but I don't.

Until now I have no idea about my data. This is my mistake I didn't consider carefully before starting.

This article is a good guidance on how to interpret qPCR data using Ct values. As people here already mentioned, normally low Ct means high expression and vice versa. Since the Ct can vary depending on parameters other than actual gene expression, you should always use known hose keeping gene/s (HKG) (one to three preferably) in order to get a clearer view on the expression of your gene of interest. If using more than one HKG you can use a geometric mean of their quantified values to get the relative expression of your genes. Hopefully this article will help.

http://www.ncbi.nlm.nih.gov/pubmed/?term=Accurate+normalization+of+real-time+quantitative+RT-PCR+data+by+geometric+averaging+of+multiple+internal+control+genes

You are welcome Trang Nguyen. Glad to help.

If the CT values of your target gene does not statistically differ from the CT values of GAPDH, all you can infer is that your gene is as much expressed as GAPDH in your samples. This, of course, if you used the same cDNA and ran both in the same plate at the same time in the same instrument and in triplicate and if GAPDH CT values does not significantly vary among your technical replicates. You can check if this is true for all your  biological replicates. The biological meaning of that information is up to you to decipher, in comparison to future assays and published data.

No, Carolina, this is simply not correct. Equal Ct values are no indication of equal transcript abundance/expression.

Thank you all,

@Arash Hamidian: I already perform this test and I don't have back-up sample to do this again. And attached file is my result.

@Carolina and Jochen: can I say that the number of copies of GAPDH and my interest gene are the same and all of their Cts < 29 so they are strong positive reactions in tumor tissue.

Would you kindly give me some ideas?

You are right. I should have said same amount of mRNA, if both probes have equal efficiency, instead of expression. Sorry about that.

 You're welcome Carolina.

Dear Carolina,

not even then (with equal efficiencies) you could conclude that similar Ct values indicate similar expression. The signal generation is sequence-specific. You may only conclude different expression levels when the ct-values are considerably different (more than 3-5 cycles, just as a house number). But having the same ct values, the expression can still be several (2-, 3-) fold different.

But I wonder what the value would be to know whether or not different genes are expressed to the same extend. Even if their relative expression levels in the cell are relevant, the post-transcriptional regulation will likely be relevant, so it would be required to evaluate the protein expression directly.

@Trang:

"can I say that the number of copies of GAPDH and my interest gene are the same"

No.

"and all of their Cts < 29 so they are strong positive reactions in tumor tissue."

I'd rather say that both genes are expressed in tumor tissue. If their expression is high or low depends on the numbers of cells from which your cDNA was used. Your ct values vary over 12 cycles, indicating a more than 2000-fold variability in the measured concentration (what is quite a lot). How much of this variabilty is attributable to differences between the amount of sample material / RNA extraction efficiency / RT efficiency and to a different cellular expression level can not be resolved with this data.

@Tran Nguyen: From what I can see in your .xlsx file, GAPDH is not a suitable HKG to use in your experiment since the Ct value variations are extreme between samples (single loading och triplicates?). Since all genes, even some general house keeping genes, are expressed differently in different conditions, it is always safe to test a battery of known HKGs and then pick the ones that are the most stable in your condition using GeNORM software. Then you can pick a stable HKG to use when measuring the relative expression of your GOI. I for instance can not use GAPDH as HKG in my experiments since its expression is affected upon hypoxic stimulation.

@Arash:

GAPDH is not meant to be a "house-keeping gene" (better: reference gene!) - the question was only if the expression levels of two genes could be compared based on their ct values.

But even if the ct values of GAPDH were intended to be used to normalize the expression of the GOI, the mere variation is no indication that it might not be useful (NB: GAPDH is surely a bad candidate for a reference gene in tumors since it is known to be regulated by hypoxia, and hypoxia plays a role in tumors - you also mentioned this). Some tumor samples are hard to get and the amount of cells used for RNA extractions can't be controlled and can vary a lot.

And regarding you comment on to "test a battery of known HKGs": this advice is generally good, but it won't help Trang in this particular case (as she wrote earlier: "Maybe my data has to give up because in my condition, re-measure is impossible.").

@Jochen: It's mean now I only can say both genes expressed in tumor cell.

I fixed amount total RNA about 400ng/reaction. So could you suggest to me the suitable method to calculate the mRNA of each gene.

Or there is no way to do with my data :)

I am sorry to say that, but there is no way.

If you prepare a standard and if you measure a dilution series under the very same conditions as the tumor samples, then you could at least give some estimate of the numbers of moelcules your ct values refer to.

But I am concerned with the large variation between your samples. If you really used 400 ng RNA/reaction, there is either some major problem with the reproducibilty of your method (extraction or RT) or both genes show a very very different expression between the samples (and this with a high correlation, what would be rather strange, at least very unexpected).

Ms. Nguyen - I would really like to try to figure out some way to help you determine something useful from your data.  Would you mind sharing a few more details about your experiment?

To start, what was the original purpose of your experiment? It doesn't seem like these data were collected to support a publication, since it is not something you can repeat. We're you looking for a trend of some sort to direct future experiments?  Learning a new technique?

I have only glanced at your spreadsheet so far.  How did you retrieve the samples, and what was the source?  In what ways are each of the samples different?  For example, are each of your data points showing tumor cells collected from a different mouse?  Or were the samples run in duplicate/triplicate?  Were any other experiments run on samples from the same source?  

@Danielle:

First time, with my limited knowledge I think very simple: all papers said this gene over-express in tumor cell, so I just want to see how about my GOI's expression in different tumor sample from different patient. And its expression relate to the existence  of cancer stem cell  in the tumor or not?

So using GAPDH is a reference gene.

I started to experiment and did not care about literature review. And after finishing I recognized my test has a lot of trouble from approaching way, to performing way. I went wrong way so now I have to suffer the consequences. This is a lesson for me.

Thank you all,

I am still not clear on one thing: are you saying that all 30 samples were collected from different patients?  And all of those patients have the cancer stem cell in their tumor, correct?  Do you have any other information about the patients...even things such as age or gender...that would allow you to group the samples?

At the risk of sounding naive or idealistic, I strongly believe that data from every  experiment should be analyzed.  Even if it has already been determined that the data will not answer your specific research question, it is a huge waste of time and money to declare that the data should be thrown away without any further consideration. Even if the method was not performed correctly, there is SOME reason that you got the specific numbers that you got.  Taking a short time to try and determine what that reason is helps you develop a strong understanding of different methods, which makes you better at designing experiments and better at interpreting more precious data. 

Dear Daniel,

Yes, all 30 samples were collected from 30 woman. I knew their old, their stage. Now I think just using this data to find the correlation between Ct value of my GOI (YB-1) and grade tumor but there is no significant statistic.

I also used Flow Cytometry method to definite the existence of cancer stem cell via their markers, and nearly all sample had CSCs phenotype.

How many did not have CSC phenotype?  If there are at least 3, you can use that as your control tissue and compare to CSC+ with ddCt method.  Also, what statistical test are you using to compare the tumor grade groups? Be careful because you cannot use a t-test to compare four groups.

If you would like, I can provide a spreadsheet to help you analyze your data.  It would be helpful to know which samples belong to each group.  However, you may want to look at your internal control (GAPDH) variability.  Were you very careful to avoid RNA degradation while you handled your samples?  Have other papers using this tissue used GAPDH as a reference gene?

Hi Danielle,

You know, my tissue is breast tumor, so I use marker CD44+, Cd24-/low to identify and nearly all sample have CSC phenotype's expression.

I used t-test to compare the difference of Ct value among 3 groups. So, would you kindly explain to me why we can not use this method, and what is the suitable test?

When I performed this test, I try to avoid RNA degradation, but looking the variance of Ct value,  I guest I can't.

I send to use my Ct data of each group to you refer.

Great!  I will make a spreadsheet as soon as I have a chance, and I will tell you more about the stats too.  What is the column "N/A"?