Standardized Fold Change

[dbaud]

Hi,

I'm new to the field of RNA-seq and my question might seem obvious to most of you, but
I'm trying to find a clear definition of the standardized fold change. As I haven't come up with anything useful off usual search engines, I hope someone has heard of this concept.
Does anyone know about the standardized fold change of a gene between 2 conditions?

[westerman]

If you are asking "what fold change is considered to be significant" then the answer is "do not use fold changes for significance; statistical methods are better but if you really want to use a fold change then go for a 3x fold change."

If you mean something else by "standardized fold change" then you will probably have to make your question more clear.

[Simon Anders]

As you are new to the field, I suppose you have come across this term and think it is a quantity in common use. To save you trouble: No, it is not. It is something specific to the paper you read it in.

A "fold change" is simply the ratio of the expressions of a gene in two conditions. I still think that "fold change" is a silly word and cannot shake off the feeling that it must have started as an explanation of the term "ratio" to a mathematically illiterate person, but now we are stuck with it.

Standardization means, in statistics, to subtract the mean or expectation, and divide by the standard deviation: http://en.wikipedia.org/wiki/Standard_score.

In some contexts, logarithms of fold changes are standardized by dividing them by their within-group standard deviation before comparing them across genes.

[westerman]

Simon's explanation is, of course, much better than my stab at an answer.

I have never considered it but I wonder if "fold change" comes from "fold coverage" which is often used similarly to "depth of coverage" in sequencing. The English definition of "fold" does not seem to imply a ratio although that is how it is used in "fold change". Ah, pondering the definitions and origins of words and phrases is sometimes a chase down a rabbit hole.

[dpryan]

I have never considered it but I wonder if "fold change" comes from "fold coverage" which is often used similarly to "depth of coverage" in sequencing.

It's just the generalization of twofold, threefold, fourfold, etc.. Fold has meant a ratio in English for centuries (my computer dictionary suggests that -fold was -fald in old english).

[dbaud]

I was refering to the fold change sometimes defined as the logarithm in base 2 of the ratio between 2 condtiions in count data (in the context of RNA-seq). That is, normalized counts are compared between 2 conditions for a given gene.
(This definition may come from analysis of Microarray data).
Then which is the correct way of standardizing the logarithm of this ratio?