I noted many discussions about SD or SEM in qPCR relative gene expression study.
Some recommended that we do not need to show SD as it does not make sense any in this analysis.
Some added SD or SEM too. Even for Calibrator (Control), although it is simply FC1 but some still show SD or SEM.
I am also confused at the end after reading all discussion and made me headache.
Because of confusion and not clearly written anywhere. I decided to share what I have found out. I found out a way to get SD for your qPCR relative gene expression study. See the attached file from a paper written by Livak et al 2001.
From there, we can noted that he use geometric mean rather than arithmetic mean. I noted that all qPCR software also use geometric mean too.
In the paper, c-myc Ct will have SD and GAPDH Ct too will have SD, although they were not mentioned there. Then Delta Ct will also have SD. This Delta Ct SD was calculated from SD of c-myc together with SD of GAPDH. It was not calculated from ave value of Delta Ct. Formula to calculate the Delta Ct's SD from the two SD mentioned above is
SD DeltaCt = {(SD of c-myc)^2 + (SD of GAPDH)^2}^1/2
We will get SD of Delta Ct and this is also followed for SD of Delta Delta Ct and finally we can get variation of FC value too according to SD of DDCt. From that 3 variation of FC (mean, upper, lower), we can get SD of FC. Would this be OK? Some said, FC should not have SD because FC is not distributed normally. So confusing. How do yo think?
For those who are having difficulties to get SD or which SD is the right one in qPCR relative gene expression for FC, this will help.
Although I have found this calculation, I am still not clear whether we should show SD or SEM or both are not necessary in this qPCR relative gene expression FC analysis.
Some of them still going on for T test or Oneway anova too. I do not understand for this too. It it so clear answer to know they are significant or not as we are looking at 1 for control. Less than 1 is down and more than 1 is up. in other words, 1 = 100%. From there we can know already clearly whether our results are significant or not. Why are some people still doing stat analysis. It is really unnecessary.
Appreciate your opinions. Thanks
The correct and most appropriate solution would be to study the kind of distribution of the data in order to calculate the SD and SEM.
lnathgubbi
I think it should be mentioned that in the paper by Livak 2001, the authors elaborate a lot on replicate reactions, i.e. you clearly have do distinguish between intra- and inter-sample variation (whatever measure you calculate for this variation...). To the end of correctly interpreting your FC results it makes sense to calculate Ct sample-wise and then sum up and analyse variation within group (treatment / control) and later on the ddCt. That would be biologically menaingful.
Best regards
Jens
Yes. It makes and sense and it is really tedious job to do. I just noticed another way of doing to get SD in his paper too. This will give answer to those who are saying that there are no SD for FC of control. By individually subtracted in order to get DDCT and take SD from that individual DDCT value, we will get all "0" for control and there is no SD for control. This make sense too because there should not be variation in standard that we refer.
We still can get SD for DDCT of treatment from variation between individual DDCT and not necessary to use formula SD= SQRT(SD1^2 + SD2^2) . This is more appropriate than the previous one if we want to present a graph for those FC comparing control vs treatment.
For me is just stunning how people use RTqPCR and analyze data since 20 years ago and the data analysis froma statistical point of view is still unclear.
New age science just suck... if referees or journals don't ask, scientist just don't care if their approach is correct or not.
Publish is the only thing important don't matter how? wow... and then we ask for money from the people taxes....
Statistical analysis of qPCR gene expression (Fold Change).
Biological replicates, Technical replicates, t-Test, pair or unpair, Two-way anova
Refer the attached paper.
Mingchuan Yu Hi! I've read questions relating to analyzing qPCR data and yet I am still confused. I would like some explanation on my own findings:
My ddCt value is -0.69 [which is down-regulation of gene], but when I apply the 2^-(-0.69), the value becomes 1.61 [which becomes up-regulation by 1.61 times].
I am very confused of why these two methods gave me different interpretation of result. I am very new to qPCR analysis and hope you can help me.
Thank you in advance.
Nur Hafizah Md Ismail A negative ddCt (if you did dCt treated - control) is actually a positive log2 fold change (log2fc = -ddCt).
So if your ddCt is -0.69, it is indeed 1.61 up-regulation (and log2fc = 0.69).
Thank you sir Valdir C Barth Jr .
So, I DO need to interpret from the 2^-ddCT value. I am confused as one of my colleague insisted on using ddCT for interpreting the result and the other choose 2^-ddCt.
Thank you for clearing up my confusion. I will use the Livak method for my study.
I think it it much more intuitive to use log2 foldchange (= -ddCt), because what is down-regulated gets a negative value and what is up a positive value, much easier for visualization. Also, the proportions of the value compared to the control are the same either if it is up or down regulated (for example, like a 4 fold change up or down are logfc 2 and -2 respectively, while in 2¨-ddCt one would be 4 and the other 0.25 --> very different numbers that may be deceiving for the human eye).
Also in one of the most used guide to qPCR, "Guide to Performing Relative Quantitation of Gene Expression Using Real-Time Quantitative PCR" (Applied Biosystems), SD is calculated as above mentioned: SD DeltaCt = {(SD of c-myc)^2 + (SD of GAPDH)^2}^1/2 . However, in this way I think that we are measuring the variability of Cts, which are not so biologically meaningful compared to the variability of dCT. Indeed we usually do statistics (e.g. ANOVA) on dCt and plot directly dCT values or Fold Change (i.e. relative expression calculated from dCT). Are you definetely agree that SD (or SEM) should be calculated directly on dCT values, rather than on Cts?
Andrea Becchimanzi I absolutely agree with you. If one of your biological replicates is more concentrated than the other, but the difference between test Ct and reference Ct are the same for both replicates, then calculating SD separately for test and reference Ct will amplify the SD of deltaCt, whereas it should be negligable. However, I am not a statistician, so I'm hoping others will weigh in...
- Badges
- Science topic
- Similar topics
- Biological Science
- Molecular Biology
- Biomolecules
- Nucleic Acids
- RNA