P-value calculation on difference in correlation
1
0
Entering edit mode
7.9 years ago
jth ▴ 190

Hi,

I have a slightly basic statistics question. I have two large correlation matrices (PCC) (one in wildtype condition, other in treatment condition) of related features. They are about 10000x10000. What i'm interested in is mainly how treatment affects the correlated nature of these features. In other words, the change in correlation. For downstream analysis, I have to use an in-house tool of our group that requires significance values (p-values). Therefore, i'm interested in the probability of a particular deltaPCC occurring due to chance.

Does it make any sense to calculate a random set of deltaPCC by selecting random features from two matrices, calculate difference of correlation from these matching or unmatching features, repeat this process for like 100000 times and calculate p-value from this distribution by looking at the percentage of generated deltaPCC's that are larger (or smaller in the other tail) than the deltaPCC of interest?

Thanks!

statistics • 1.9k views
ADD COMMENT
0
Entering edit mode

I'd ask CrossValidated if I were you.

ADD REPLY
3
Entering edit mode
7.9 years ago

You can use the Mantel test if the two matrices are independent. The Mantel test measures the correlation between two distance/similarity matrices and gives an associated p-value. The test can be found in several R packages: ape, ade4, vegan.

ADD COMMENT
0
Entering edit mode

Looks interesting! Thanks a lot!

ADD REPLY

Login before adding your answer.

Traffic: 2217 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6