statistical test for interference of correlation

Question

[markus.glass]

Hi,

There is a factor A that is known to stabilize another factor B on the RNA level. This can be seen e.g. by a nice positive correlation of both factors on the RNA expression levels (of course, this would not be sufficient to conclude the stabilization theory but there is plenty of other evidence for it).

Now, there is a factor C that is hypothesized to interfere with the binding of A and B and thereby disturbing the stabilization of B via A.

Now my question, does anyone knows an appropriate stat. test of testing this hypothesis?

In other words, I have RNA expression values of all 3 factors in a large set of samples and I want to know if high expression of C interferes with the correlation between A and B.

Answer 1

[LChart]

This is precisely the setup of 'Liquid Association' (https://www.pnas.org/doi/10.1073/pnas.252466999, https://onlinelibrary.wiley.com/doi/abs/10.1111/j.1541-0420.2010.01440.x, https://www.bioconductor.org/packages/release/bioc/html/fastLiquidAssociation.html).

You could also use a linear model with an interaction term:

A ~ 1 + B + C + B:C

and ask if the coefficient on B:C is significant.

[markus.glass]

Hi LChart,

Liquid association indeed seems to be exactly what I'm searching for. Thanks!