help me(one-way or two-way ANOVA??)

5 posts / 0 new
Last post
Eunah
Eunah's picture
help me(one-way or two-way ANOVA??)

Hi scientists~

I am a graduate student. Now I am doing ChIP with lymphocyte cell line derived from human.
so, I have duplicate ChIP data for each cell. Total number of data for 1 gene are going to be 12 because I have 3 control cells and 3 patient cells.
Since ChIP is performed to get certain protein occupancy at one gene with several primers that recognize several region of the gene, I have many data, and I am frustrated to get 'p value'.
Q is that what kind of statistics should I use?

Summary..
I have 4 primer set that recognize 4 different region of my target gene (e.g. promoter, exon1, exon2, and exon3) in both 3 control cells and 3 patient cells. All experiment are duplicate..
I am thinking that I could use two-way ANOVA because I have two factors that can affect protein occupancy at my target gene ( control vs patient, dependant on region of target gene).
Moreover, I should show significance of protein occupancy between control and patient at certain region of target gene. how can I get it? I am thinking I should make one average bar out of 3 control cels and one bar out of 3 patient cells and do T-test??
Or..I am using Prizm two-way ANOVA and posttest option to get p-value b/w control and patient...

Am I right?
Or is there anybody who can help me?

R Bishop
R Bishop's picture
I think I get your question,

I think I get your question, so as a first crack at your problem. we might have to go back and forth a bit to get it.

from the Prizm manual - "Two-way ANOVA, also called two-factor ANOVA, determines how a response is affected by two factors. For example, you might measure a response to three different drugs in both men and women. Drug treatment is one factor and gender is the other. "

I dont think you have two variables. Just one, a second variable would be temperature for example.

 I believe based off what you told us you have it right. You should be able to determine the mean for occupation of 1 site on your gene (promoter, exon 1, etc) in either your 3 control or 3 patient cells and compare those data by a t-test in Prizm. Thus you could directly compare occupation of the promoter in 1 gene in control or patient cells.

Here's where I may be wrong, its hard to tell without knowing a lot more about your experiment.  I also think you can determine the mean for each location on your gene in triplicate, enter those 4 values into a single results column for the total gene under either control or patient and, determine the mean occupation of the gene in total.  Then use that data to generate a p-value based off a T-test using Prizm as well. 

Somebody check me on that if you are reading this. 

Bishop

Eunah
Eunah's picture
Thanks Bishop,

Thanks Bishop,

Actually, I was thinking about your last comment..put a mean value out of triplicate into one cell in the Prizm..I can make one column for 4 different regions (making rows) of my target gene for control and patient..

However, I have 3 control cells..and 3 patient cells..
For one group, I need 3 sub-columns..I have not found t-test for groups yet..
To use t-test..I should put a mean of mean of 3 control cells..that is what I got so far..but it is not good right? mean of mean??? I am supposed to put real data into cells of statistics software right? and then I should let software do their job like getting mean, SD, and whatever out of my real data not mean...
I know once I have a mean that already reflect my real data..but still I have variations between triplcate..how can I deal with that??

anybody help me??

R Bishop
R Bishop's picture
Hey Eunah,

Hey Eunah,

Have a look at this blog

http://www.workingthebench.com/2007/08/qpcr-analysis-of-chip-data.html

Perhaps this is the explanation for how to deal with the error in your samples.

Im not giving up, just somethin i stumbled on.

Bishop

Eunah
Eunah's picture
Thanks Bishop again...

Thanks Bishop again...
I saw the blow..it is good...but i need more statistics skill beyound the blog..
I already have ddCT from 3  control cells and 3 patient cells.
I should have 'p-value' b/w controls and patients.
Can I put mean value out of 3 controls and 3 patient into one cell of Prizm to do t-test??
That is mean of mean, right??
Most scientist do that??