4 replies to this topic

[toejam]

hi guys,

this is the first time that i'm i'm currently analising microarray data, so i'm not fully familiar with it and i'm trying to get the grips on it.

i've found some differentially expressed genes on metabolic pathways that i'm interested using first robin and then tranferring that data into mapman, some other genes of interest i've found them manually because the pathway i'm interested in is not predesigned in mapman.

anyways, i know it is a log2 value in the fold change of the expression of the genes, but some of these values are negative.

in order to get the fold change, say for example, if the logFC value is 2, then the absolute fold change would be 4 times the expression level of one treatment over the other one (2^2), right? but as i said, there are values of -2 or less, and if i do 2^-2 the value is 0.25 and i'm having a bit of trouble understanding what that exactly means...

in my head, if the absolute fold change is 0.25, or even lower if the negative values are larger, then it might not be significant?

thanks in advance.

tj.

[nanook]

positive logFC indicate the logarithmic foldness of UPregulation
negative logFC indicate the logarithmic foldness of DOWNregulation

logFC = log(expr1)-log(expr2)
e.g.
log2FC = log2(expression in mutant backround) - log2(expression in wildtype)

if a gene has an expr level of 16 in wildtype and an expr level of 4 in the mutant
the log2FC is -2

[toejam]

thanks nanook, thing is that i was swapping between scales, then it got confusing. but now it's clear, or at least i would like to believe so

cheers!

[honey]

I have a custom array gene expression data with controls and Cancer as log transformed RMA normalized values. If I have to calculate fold chnage or difference in cancer and normal tissues -
I will simply take difference of values cancer - normal and it will be log fold change and then if I want in linear scale I can take antilog of this difference.
Could some one advice if I am correct or I am doing something wrong. Thanks.

[nanook]

honey, on 18 January 2012 - 10:51 AM, said:

I have a custom array gene expression data with controls and Cancer as log transformed RMA normalized values. If I have to calculate fold chnage or difference in cancer and normal tissues -
I will simply take difference of values cancer - normal and it will be log fold change and then if I want in linear scale I can take antilog of this difference.
Could some one advice if I am correct or I am doing something wrong. Thanks.

exactly as you said!