How to assess the Fold-change significance using a t-test in R?

Question

[lmy941003]

Hello,everyone!

I want to assess the significance of Fold-change using t-test. Normally, if I know the data of case and control, I can perform the t-test, but the problem is that I only have the log2 fold change ,so I don't know how to performing the t-test only by calculated values of fold-change. Is it possible?

I am new to this field, please help. Thank you very much!

[Devon Ryan]

Do you have the SE of the fold-change? You can't calculate the p-value from only a fold-change, you'll need some other value to let you know how confident that estimate is.

[Carlo Yague]

To the best of my knowledge, this is not possible with just the fold change.

Answer 1

[dariober]

Granted that with just one data-point you can't do much without making further assumptions. It seems to me that you want to know how extreme your log-fold change is. For this, you could make a guess, more or less driven by data you have, about the distribution of log-fold changes from which your data-point comes from. Then, you can ask where your log-fold-change maps on this distribution. For example, if you assume the distribution is normally distributed with mean 0 and standard deviation 2, then (in R) pnorm(q, mean= 0, sd= 2) tells you how extreme your data point q is. For example:

pnorm(q= 2.5, mean= 0, sd= 2) # 0.89, not very close to 0 or 1, not very extreme
pnorm(q= 5, mean= 0, sd= 2) # 0.99, quite extreme
pnorm(q= 5, mean= 0, sd= 5) # 0.84, not very extreme anymore