Comparison of microarray and RNA-seq data

Question

[bharata1803]

Hello,

So, currently I want to do something like a comparison of micro array and RNA-seq data. In micro array, I have sample A and B1 and in RNA-seq, I have sample A and B2. I want to know the difference between B1 and B2 in terms of changes that occur from A. My method is list all DEG from microarray and check it in RNA-seq data. If the result is different (for example, in micro array it is up reg, but in RNA-seq it is not upreg) I will think of it as the difference.

My question is, is the comparison acceptable? How to check statistically the difference that occur from this comparison is biologically significant?

Thank you for your answer.

Answer 1

[anioannides6]

I believe that the comparison of microarray and RNA-seq is not totally acceptable. In microarrays the signal, even when it's interpreted as value, has a max threshold that cannot surpass. It's an artifact due to the technique itself, because the microarray platforms can only measure up to a certain amount of signal, ie if one gene has an expression of let's say "6" and an other has "10" the two signals will be interpreted the same (don't know exact values, just giving an example). To check for statistical significance you can use R or Prism (for smaller datasets) and perform a t-test.

[bharata1803]

Thank you for your answer.

So, what about if it is not directly comparing the value. It just see the pattern. For example, I want to see whether the comparison I make in micro array is upregulated or not and check it to the sample with RNA-seq. I want to check the possibilities if in micro array up reg, for example, is in RNA-seq also up-reg, down-reg or no difference. So, I think I will not compared the values directly. With this method, I will have a list of DEG from microarray and list of RNA-seq genes that doesn't have the same pattern with result from micro array.

What I'm still in doubt is the meaning of the fold change. Does it have the same meaning of fold change in terms of biological meaning?

[anioannides6]

I guess if you are to just check the pattern, it will be ok. If you are not interested in comparing the values directly maybe a heatmap will help you observe changes and similarities, although this is just a rough estimation. If you want to be more accurate I suppose you will have to perform statistical analysis between A-B1 and A-B2. Fold change describes how much a quantity changes from it's initial value, ie if a gene in B1 presents 2 fold change it means it's expression is 2 times greater or smaller (depending on the final value) from the initial value in A.