6 replies to this topic

[gyma]

hi I am working on 1 gene with several isoforms. To determine the abundance of different isoforms of this gene in various cell lines, I designed two sets of primers that are each only specific to certain isoforms .

After I got the qPCR result, I found it obscure and started to question this method. Is this method ok? I thought it would be similar as if I compare 1 gene with GAPDH. In this case, by comparing their CT value, I think I could know the relative abundance of the gene, e.g. 1/1000 of GAPDH. Am I right? Or this method is incorrect because one cannot compare between different genes? Please give me some advice. Thank you very much.

[Trof]

You can relatively compare two genes by normalizing one to the other (yes, like to a reference gene). Let's say you want to compare relative abundance of isoform 2 to isoform 1. You just fill in the equation as you would for reference gene and target gene.
But if you have different expression of isoform 1 between your samples, this information will be lost, because the isoform 1 is taken as a "stable" reference among all samples. You need to decide whether you need this kind of information or not.

If you want to see changes in transcription within your samples in both isoforms you need to actually do relative quantification to a "real" reference gene for both of them, and since then they will be both normalised to same reference you could use that also to compare relatively one isoform to the other.

Other option is to do an absolute quatification comparing both to a standard, which will give you absolute values that can be then normalised to whatever and will also tell you the actual abundance of each isoform. For this is of course better to make a plasmid standards (diluted with dummy NA), but if you only testing one experiment and don't need to compare to other experiments, PCR-derived qPCR standards could be freshly made and used.

[gyma]

Trof, on 06 December 2012 - 04:47 AM, said:

You can relatively compare two genes by normalizing one to the other (yes, like to a reference gene). Let's say you want to compare relative abundance of isoform 2 to isoform 1. You just fill in the equation as you would for reference gene and target gene.
But if you have different expression of isoform 1 between your samples, this information will be lost, because the isoform 1 is taken as a "stable" reference among all samples. You need to decide whether you need this kind of information or not.

If you want to see changes in transcription within your samples in both isoforms you need to actually do relative quantification to a "real" reference gene for both of them, and since then they will be both normalised to same reference you could use that also to compare relatively one isoform to the other.

Other option is to do an absolute quatification comparing both to a standard, which will give you absolute values that can be then normalised to whatever and will also tell you the actual abundance of each isoform. For this is of course better to make a plasmid standards (diluted with dummy NA), but if you only testing one experiment and don't need to compare to other experiments, PCR-derived qPCR standards could be freshly made and used.

Thank you so much for such a quick reply. Regarding the part in red, I am confused. Lets think of it in this way, it is not about isoforms, I want to compare the abundance of 2 different genes. In this case, I can get the expression pattern of gene A or gene B in my samples. And according to what you said, I can know the relative abundance of A and B in each sample by normalizing B to A. By doing so, I think all the data is comparable. Is it not correct that I put all data in one figure?

[Trof]

Relative quantitation uses one gene as a normalisation factor for the other gene, that way slight differences between the actual amounts of template can be eliminated (in theory).
Identically you can use it to make a ratio of any other genes, as long as you are only interested about relative differences between them. And of course only relatively to some calibrator or control sample = comparing.

To illustrate it lets say you get these absolute values (in ng or milions of copies to make it easy):

Sample 1 - gene A = 10; gene B = 100  (calibrator sample)
Sample 2 - gene A = 30; gene B = 50
(and lets say some reference gene R is same for both and equal 1)

In absolute quatification, you would get these numbers, you would just divide them to get various ratios. Equation to do this would be for gene A: (gene A of sample 2/gene R of sample 2)/(gene gene A of sample 1/ gene R of sample 1)

Relative quatification can only give you certain ratios, ratio of gene A in sample 2 normalised to reference and sample 1 = 3-fold increase
ratio of gene B in sample 2 normalised to reference and sample 1 = 2-fold decrease  (or 0.5 if you like)
and with the same data you can normalise gene B to gene A within samples 1 and 2 by dividing the same way (just instead of gene R you would divide by gene A) = 6-fold decrease (or 0.17) of B relative to A in sample 2 normalised to sample 1.
It won't tell you for example how many times is gene A more or less abundant than gene B in sample 1. Because without double-comparison relative quantification doesn't work.

Just you need to realize, that in relative quantification you don't have the numbers (of copies, of ng or of anything) just Ct values that are powers to efficiency, hence you would use the delta-delta or Pffafl equations to get the same result as you would get by normal double division of the absolute values by substracting Ct diferencing and then powering that.

Point is, absolute values can be used to produce relative ratios, the oposite is not possible. But if you can do with ratio of A and B genes separately, and with a ratio of B to A, you can use it (but include reference gene). If you want to know ratios of genes within single sample, you need absolute quantification.

[gyma]

Trof, on 06 December 2012 - 08:20 AM, said:

Relative quantitation uses one gene as a normalisation factor for the other gene, that way slight differences between the actual amounts of template can be eliminated (in theory).
Identically you can use it to make a ratio of any other genes, as long as you are only interested about relative differences between them. And of course only relatively to some calibrator or control sample = comparing.

To illustrate it lets say you get these absolute values (in ng or milions of copies to make it easy):

Sample 1 - gene A = 10; gene B = 100  (calibrator sample)
Sample 2 - gene A = 30; gene B = 50
(and lets say some reference gene R is same for both and equal 1)

In absolute quatification, you would get these numbers, you would just divide them to get various ratios. Equation to do this would be for gene A: (gene A of sample 2/gene R of sample 2)/(gene gene A of sample 1/ gene R of sample 1)

Relative quatification can only give you certain ratios, ratio of gene A in sample 2 normalised to reference and sample 1 = 3-fold increase
ratio of gene B in sample 2 normalised to reference and sample 1 = 2-fold decrease  (or 0.5 if you like)
and with the same data you can normalise gene B to gene A within samples 1 and 2 by dividing the same way (just instead of gene R you would divide by gene A) = 6-fold decrease (or 0.17) of B relative to A in sample 2 normalised to sample 1.
It won't tell you for example how many times is gene A more or less abundant than gene B in sample 1. Because without double-comparison relative quantification doesn't work.

Just you need to realize, that in relative quantification you don't have the numbers (of copies, of ng or of anything) just Ct values that are powers to efficiency, hence you would use the delta-delta or Pffafl equations to get the same result as you would get by normal double division of the absolute values by substracting Ct diferencing and then powering that.

Point is, absolute values can be used to produce relative ratios, the oposite is not possible. But if you can do with ratio of A and B genes separately, and with a ratio of B to A, you can use it (but include reference gene). If you want to know ratios of genes within single sample, you need absolute quantification.

Thank you very much for such a detailed explanation. I tried to reply immediately but then I found I couldnt because I got more confused. Now I am still not so clear about this problem but I tend to believe that we cannot compare the relative abundancy of different genes in one cell without absolute quantification. When using the relative quantificatioin method such as qPCR with SYBR green, in the case of comparison of 1 gene in different cells, theoretically, the only factor that affects the result is the abundency of the gene in cells. However, in the case of comparison of gene A to gene B, there are more than one factor. For example, the primers will be different, and the reaction efficiency will also be different. When you see gene A with a bigger Ct value, you might think it is less in cells than gene B. But what if this gene is somewhat harder to amplify (sequence issue or others) so if you do the absolute quantification you might find they have same abundancy.
Another reason that makes me believe that we cannot do such comparison is this. Imagine we are doing relative quantification on 1 single gene with two primer sets. The product sizes of these two primer sets are different, say 200 and 100 bp each. If we use Taqman method, I think the 200 one will take more time to reach the threshold so it will result in a difference in the Ct value. But that doesnt mean anything because this is the same gene. However, if we use SYBR green method, although 200 bp takes more time, I guess it also generates more signals due to longer length. I dont know what the result will be but they could be different from the Taqman method. So now I think without guaranteeing that you have the comparison based on the same background, it is incorrect to do so.
anyway, this is my opinion. I really look forward to your response. Thanks for your time.

[Trof]

Yes within on cell(s), you would need absolute quatification. If you are comparing transcipt A to levels of transcript B.
But if you want to know how transcript A levels change within different cell lines (compared to one of them), and the same for transcript B separately, you can use relative quantification.

The problem with different efficiency (and product length, which transforms into efficiency) can be corrected in relative quatification (in absolute quatification it is corrected also, since you use standards) by using Pfafflequation that can calculate with different efficiencies of target and reference gene(s). There are other opinions that the assays should be ideally all around 100% so, then all the efficiencies would be very similar and the correction doesn't really make it any more accurate, since the efficiency measurement itself has a slight error.

[gyma]

Trof, on 18 December 2012 - 02:22 AM, said:

Yes within on cell(s), you would need absolute quatification. If you are comparing transcipt A to levels of transcript B.
But if you want to know how transcript A levels change within different cell lines (compared to one of them), and the same for transcript B separately, you can use relative quantification.

The problem with different efficiency (and product length, which transforms into efficiency) can be corrected in relative quatification (in absolute quatification it is corrected also, since you use standards) by using Pfafflequation that can calculate with different efficiencies of target and reference gene(s). There are other opinions that the assays should be ideally all around 100% so, then all the efficiencies would be very similar and the correction doesn't really make it any more accurate, since the efficiency measurement itself has a slight error.

thank you very much.