Lineal mutual information for ensemble NMA

[tasarte]

Hi,

I would be interested in comparing the lineal mutual information coefficients for a set of structures, which are divided into two groups. I have already prepared an ensemble NMA, and there are clear dynamic differences in the fluctuations between the two groups. However, as far as I can see, only cross correlation is available for ensemble NMA. Is this correct, or did I miss something in the documentation?

Many thanks!

‌

‌

‌

[Xinqiu Yao]

Hi,

‌

You can use the cov() function to calculate the covariance and then use the cov2dccm() function with method="lmi" to get LMI.

[tasarte]

Hi Xinqiu,

Thanks for your answer. I’ve tried to use the cov() function as shown in the documentation but it’s not working as it should.

> class(test)
[1] "enma"

# As shown in the docs for ensemble NMA
> cov_matrix <- cov(test, ncore=NULL)
Error in cov(test, ncore = NULL) : unused argument (ncore = NULL)

# As shown for single structure NMA
> cov_matrix <- cov(test)
Error in cov(test) : supply both 'x' and 'y' or a matrix-like 'x'

Any ideas of what the problem could be? For reference, I am using bio3d v2.4-4.

‌

‌

[Xinqiu Yao]

Can you try cov.enma() directly and see if it works? Also, if you want, you can attach an example “test” file for me to take a deeper look.

[tasarte]

Hi Xinqiu,

Indeed, cov.enma() works. How do you go about using cov2dccm() since it’s not exported to the namespace? I’ve tried some variations around bio3d:::dccm.xyz.cov2dccm(cov_matrix) but I didn’t find what works. Apologies, I don’t use R regularly. This could be an addition to the docs.

‌

[Xinqiu Yao]

it should be bio3d:::.cov2dccm(), note that there is a period “.” before “cov”.

[tasarte]

Brilliant, I got the linear mutual information matrix for each structure. Thanks a lot for your help!