I want to do research to compare mRNA in two groups and want to calculate the sample size for the study. Most of the articles has given expression of mRNA as fold change. Please help me how can i use these values, and what will be the formula.
Do the other articles report the fold-change of the same mRNA you are going to study?
If not, then their values will not be very useful for you.
If yes, the these values could be used as "resonable effect sizes" you should aim to detect with sufficient precision in your study. What you will need additionally is an estimate for the precision, too (i.e. how variable will the results be from different samples/individuals). If this is not given in the articles, you might either need a pilot study or you need to make an educated guess.
This is the procedure:
1) express the effect d as the LOG fold-change
2) get the standard deviation s of such log fold-changes from different samples. An educated guess is that s is about log(2) to log(5).
3) for a reasonable precision your study will then need about n = 16*s²/d² individuals.
Example: consider the folg-change is 2. So d = log2(2) = 1, and let s = log2(4) = 2, hence n = 16*s²/d² = 16*2²/1² = 16*4 = 64. Note: the log2-values refer directly to Ct-alues from qrtPCR. This means: d is nothing but the ddCt, and s is the standard deviation of dCt values!
See: STRUTS:Statistical Rules of Thumb, Chapter 2, page 3:
www.nrcse.washington.edu/research/struts/chapter2.pdf
Why use log fold-chage? - Because the distribution of fold-changes is roughly log-normal, so the distribution of log fold-changes is roughly normal, and the standard analyses (e.g. using the mean as central value, all the standard linear model stuff, including the rule of thumb cited above) are all based on the assumption that the distribution is approximately normal.
Hi,
There are two ways to get fold change.
1) Using rules by making your own code using programs such as R
2) Log your RNA expression values first and then divide them.
Zaima,
Can you clarify what you mean by "sample size"?
And some ways of calculating fold change (or fold difference) are attached here.
Do the other articles report the fold-change of the same mRNA you are going to study?
If not, then their values will not be very useful for you.
If yes, the these values could be used as "resonable effect sizes" you should aim to detect with sufficient precision in your study. What you will need additionally is an estimate for the precision, too (i.e. how variable will the results be from different samples/individuals). If this is not given in the articles, you might either need a pilot study or you need to make an educated guess.
This is the procedure:
1) express the effect d as the LOG fold-change
2) get the standard deviation s of such log fold-changes from different samples. An educated guess is that s is about log(2) to log(5).
3) for a reasonable precision your study will then need about n = 16*s²/d² individuals.
Example: consider the folg-change is 2. So d = log2(2) = 1, and let s = log2(4) = 2, hence n = 16*s²/d² = 16*2²/1² = 16*4 = 64. Note: the log2-values refer directly to Ct-alues from qrtPCR. This means: d is nothing but the ddCt, and s is the standard deviation of dCt values!
See: STRUTS:Statistical Rules of Thumb, Chapter 2, page 3:
www.nrcse.washington.edu/research/struts/chapter2.pdf
Why use log fold-chage? - Because the distribution of fold-changes is roughly log-normal, so the distribution of log fold-changes is roughly normal, and the standard analyses (e.g. using the mean as central value, all the standard linear model stuff, including the rule of thumb cited above) are all based on the assumption that the distribution is approximately normal.
- Badges
- Science topic
- Similar topics
- Biological Science
- Molecular Biology
- Molecular Biological Techniques