Statistical analysis gene expression correlation

[RobbenStijn]

Dear all,

I've got a zinc finger protein (znf479) that binds on 920 genes in the brain. We hypothesize that the binding of this protein on a gene inhibits the transcription of these genes.

To prove this we checked the correlation of the expression of these genes with the expression of the zinc finger protein. We found that around 62% of the expression of the genes that znf479 binds on is negatively correlated with the expression of znf479.
This is interesting because when we looked at the distribution of all expressional correlations of the genes in the brain is was around 60% POSITIVE and 40% NEGATIVE.

We now want to use a statistical test to check if this is just coincidence, or that it is possible to say that znf479 inhibits gene expression to some extend. We obtained all the correlation values from brainspan.com.

Any help would be greatly appreciated!

Cheers,

Stijn

[Richard Finney]

Well.

What's the number of genes used in the 60/40 ratio calculation of "other" genes?

NB: correlation != causation. Even if you get a p-value, it's not "proof".

You'll need to do careful experiments to provide convincing evidence that your zinc finger is doing interesting things.

[RobbenStijn]

The number of "other" genes is around 60000 (all from the brainspan data). Would you recommend a statistical test?

I know it's not proof, but an indication would suffice. Thank you very much for your help.