Log2 transformation - ECL or radioactive detection and Log2 scale (Oct/30/2008 )

Dear All,

I quite wonder about the microarray data. Since I read from any book, they mentioned about the fluorescent dye, after the array was scanned the data of each spot must be transformed in to Log2 scale.

But, if we use the radioactive or ECL as a detection system, is it necessary the transform the array data into Log2 scale? Why?

Anyone who has the experience in this please help me to clearify this question.

---

My understanding is that log2 transformation causes the raw data set to be more normal without changing the relationship between the values. This is useful for many reasons, not the least of which is because many statistical tests assume a normal distribution of values. Assuming my understanding is correct, the matter in which the signal values were generated is irrelevant.

---

Thank you for your kind explainatin.

From your mention, the array signal should be transformed although, it come from ECL or radioisotope, right?

If get the wrong message that you would like to describe me, please let me khow.

Thank you very much.

---

Hello,

Maybe, it would be a bit late to answer !
Anyway, log transformation is simply to reduce the values to get them smaller to strech down on graph axes.
For exemple, for a hybridization value of 10 000 pixels (hybridization values are often higher than this!), you will need a long "X" axe to point it on, while log2(10 000) will contract it into only (13,3), which will be easily graphed in.
Now, imagine you have a value of a much higher than 10 000 ? it will be difficult to deal with without log transformation.
This is why, log10 is also sometimes used to reduce vlaues when they are very high.

---