Variance - basic statistics

[sazz]

Hello all,

I'm sorry for my very naive and basic question, but I am trying to understand a couple of graphs: dispersion, M vs A etc, and I am a little confused about "variance" term.

When you check the formula of variance it is the average of the squared differences from the mean. So can I say, genes with high FPKM values tend to have "high variance" and also they are more dispersed relative to low expressed genes? (But I guess, high variance in high FPKM is not a problem when you plot a negative binomial distribution graph to calculate the significance of differential expression) But this also sounds odd because without thinking the math part, I am tempted to say, low expressed genes generally are not significant in differential expression analyses due to the "variability" between FPKM values I guess, there is a misconception here (for me) as I think the variability in "percentage". Moreover, this variability defines the shape of the negative binomial distribution, if it will be more squeezed or spread, used for statistical testing, right? :/

Sorry for asking about basic statistics. I would appreciate if one could explain briefly.

Thanks!

[csmatyi]

Hello everyone, I have a statistics question:

I have a data set:

1
1
1
0
1
1
0
1
0
0
0
0
0
0
1
0
0

I want to measure how well the 1's accrue at the top of the list, that is, how well the 1's and 0's separate. The above list should have a high value compoared to a random one:

1
0
1
0
1
0
1
0
...

What kind of test do I need for this?

Thanks!

[TiborNagy]

Hi sazz!

Variance in gene expression is not depends on FPKM. It is just a statistical measure of the replicates.

csmatyi:

I think you need khi-square test.