发信人: Subaoma (X5 Driver), 信区: Biology
标 题: Re: Question of microarray:gene expression
发信站: Unknown Space - 未名空间 (Fri Sep 3 12:46:35 2004) WWW-POST

That's why they do normalization. Actually there is always variability between
two microarrays. Even you use same amount of RNA from the same biological
sample, preparation and hybridization will definitely bring in bias. Because
there are too many steps involved and it's impossible to make them same
between any two arrays: reverse transcription, labeling, hybridization,
washing, scanning......So they use housekeeping control to normalize sample
preparations and spike-in control to normalize hybridizations. Besides, they
developed many mathematic models to do multi-arrays normalization. For
example, mean expression level of all genes on each array should be the same;
or, expression leves of all genes on each array should fit a same curve.

Anyway, microarray just provides a global and rough picture of mRNA pool. It's
not a sensitive method especially for low expressed genes. You can not publish
a microarray paper without confirmation by other gene expression methods or
protein assay.



【 在 tootootootoo (tutu) 的大作中提到: 】
: I read some papers regarding using microarray to evaluate drug effects on
gene
: expression. They all choose to use same amount of RNA from both control and
: study groups to evaluate effects. Anyone knows why?
:
: As what I understand, if a gene DNA is overexpressed, it should have more
: amounts of mRNA than that of the control sample.
:
: For example, there are genes of DNA_A and DNA_B. In control group without
drug
: treatment, they have corresponding mRNA_A 50 and mRNA_B 50(total 100). In
: study group, if a drug induce the DNA_A, say, double mRNA to 100, and the
: mRNA_B is still 50 (total 150).
:
: According to papers, we need to take out same amount (say, 100) of total
mRNA
: from both groups and then to apply to microarray chips.
:
: But in this way, those from the study group will be 2/3 of mRNA_A(67) plus
2/3
: of mRNA_B(34). and we should find a little bit overexpressed mRNA_A and
: less-expressed mRNA_B. Am I right?
:
: Obviously if above is true, it would be not a good design.
:
: So please, anyone knows why, please give me some hints about it!
: (Actually same method also used by western-blot evaluation of
: gene-expression.)
:
: Thanks in advance!!!
:
: tutu
:
:

--
BMW, 我的最爱！
※ 来源:．Unknown Space - 未名空间 mitbbs.com．[FROM: 170.140.]