I am using the Bio-Rad CFX Manager Gene Study tool as my GeNorm interface, but I am not sure how to determine the optimal number of reference genes to use by calculating the V value for V2/3, V3/4, etc... This tool does not directly give me the V values.
Does anyone have experience with the Bio-Rad Gene Study tool and know how to calculate the V values with it?
Alternatively, can someone explain to me how the V value is actually calculated? I know it is done via step-wise inclusion of less stable reference genes to the most stable pair, but how is V actually obtained? For example,, for V2/3, Is it a comparison of the geomean M value when the two most stable genes is used, to that when the three most stable genes are used? And if so, what kind of comparison? Difference? Percentage? I am kind of lost in regards to this, so any help would be appreciated.
Thanks in advance!
I did the calculation of M and V values on my own in excel, so I had to figure these things out. I can't help you with your BioRad program, but the calculations for the V value are really easy and can easily be computed in excel from a list of samples of a particular type and the CT values for your endogenous controls. Calculation of M value in excel takes a bit more work but its doable. The interpretation of the V value is as John mentioned...
The method for calculation of the pairwise variation Vn/n+1 is just like the calculation of M value. As you know when you calculate the M value, you exclude the worst endogenous control the recalculate, so if you had 8 endogenous controls to start and you found that B actin had the worst M value, you would exclude it. The V7/8 value then tells you how much variation in the normalization factor (geometric mean of endogenous controls) is attributable to B actin. So for this example to generate the V7/8 value (comparing normalizing with all 8 controls or 8 - B actin =7 ), you calculate normalization factors (NF) for every sample when you use 8 or 7 endogenous controls. So when you use 8 controls, NF8 = geometric mean of all 8 endogenous controls, and NF7 = geomean of the remaining 7 endogenous controls excluding B actin. So now you have a list of values for NF8 and a list of values for NF7. Then for each sample you calculate the ratio between NF8 and NF7 (which is log transformed to normalize the distribution about 0), this value is called A7/8 and A7/8=log2(NF7/NF8). So now you have a list of A7/8 values and each value corresponds to one sample. Then from that list of you calculate V7/8 which is just the standard deviation of the data set A7/8.
TLDR:
A7/8 = log2(NF7/NF8)
I believe V2/3 represents the pairwise variation of two genes compared to that with three, V3/4 represents three compared with four, and so on. It often forms a sort of U shape as going from 2 to 3 to 4 often represents a slight improvement, but beyond that you're including poorly scoring genes so the variation (and thus V) increases.
Basically though: try and get a value below 0.15.
If you're in this ballpark with two genes, and are getting only a minor improvement going from 2/3 to 3/4, then stick with two. Three is probably the most you should ever NEED, if you've picked your candidates well.
If it's looking like you need four genes, or you're above around 0.2 no matter what combination of genes you use...then probably pick different genes.
You can check the exact formula on the vandesomple article from genome biology
"accurate normalization of real-time ..."
I did the calculation of M and V values on my own in excel, so I had to figure these things out. I can't help you with your BioRad program, but the calculations for the V value are really easy and can easily be computed in excel from a list of samples of a particular type and the CT values for your endogenous controls. Calculation of M value in excel takes a bit more work but its doable. The interpretation of the V value is as John mentioned...
The method for calculation of the pairwise variation Vn/n+1 is just like the calculation of M value. As you know when you calculate the M value, you exclude the worst endogenous control the recalculate, so if you had 8 endogenous controls to start and you found that B actin had the worst M value, you would exclude it. The V7/8 value then tells you how much variation in the normalization factor (geometric mean of endogenous controls) is attributable to B actin. So for this example to generate the V7/8 value (comparing normalizing with all 8 controls or 8 - B actin =7 ), you calculate normalization factors (NF) for every sample when you use 8 or 7 endogenous controls. So when you use 8 controls, NF8 = geometric mean of all 8 endogenous controls, and NF7 = geomean of the remaining 7 endogenous controls excluding B actin. So now you have a list of values for NF8 and a list of values for NF7. Then for each sample you calculate the ratio between NF8 and NF7 (which is log transformed to normalize the distribution about 0), this value is called A7/8 and A7/8=log2(NF7/NF8). So now you have a list of A7/8 values and each value corresponds to one sample. Then from that list of you calculate V7/8 which is just the standard deviation of the data set A7/8.
TLDR:
A7/8 = log2(NF7/NF8)
Just for everyone's information:
If you buy the BioRad CFX Manager software, you get a free subscription for qBase+ software which allows you to calculate M and V values simply by importing your run data.
I figured this out, so I just went in this direction. Thank you for your answers.
- Badges
- Science topic