About writing NMD file

[TianmingQu]

Dear ProDy developer,

Is there anyway I can write an NMD file with the PCA vector or displacement I got from other resources instead of ProDy so that I can visualize it on VMD?

Best, Tianming

[jamesmkrieger]

Hi Tianming,

I'd say the answer is yes. You just need to get your PCA or displacement vectors into ProDy format.

For example,

my_pca = PCA()
my_pca.setEigens(my_vectors, my_variances)

or

my_defvec = Vector(my_vector)

[TianmingQu]

Thanks for the quick response James! Just to clarify, after my_pca.setEigens(my_vectors, my_variances), my_pca would be the ‘mode’ input for writeNMD, is this correct?

Tianming

From: James Krieger @.> Date: Wednesday, February 8, 2023 at 11:59 AM To: prody/ProDy @.> Cc: Tianming Qu @.>, Author @.> Subject: Re: [prody/ProDy] About writing NMD file (Issue #1656)

Hi Tianming,

I'd say the answer is yes. You just need to get your PCA or displacement vectors into ProDy format. For example,

my_pca = PCA()

my_pca.setEigens(my_vectors, my_variances)

or

my_defvec = Vector(my_vector)

— Reply to this email directly, view it on GitHubhttps://urldefense.com/v3/__https:/github.com/prody/ProDy/issues/1656*issuecomment-1422944038__;Iw!!Epnw_ITfSMW4!safYNzJQqBD5AotifrXZDpm6Nsefz1NyCAW8p0I3uMTYDuxoMpl0cTHOUc7ewo4Ss1KuIZrs4i17pcNcKcRtZ8TrLfE$, or unsubscribehttps://urldefense.com/v3/__https:/github.com/notifications/unsubscribe-auth/AQXNBSTDGGA4J27WI5U6T5DWWPGHHANCNFSM6AAAAAAUVPYJFY__;!!Epnw_ITfSMW4!safYNzJQqBD5AotifrXZDpm6Nsefz1NyCAW8p0I3uMTYDuxoMpl0cTHOUc7ewo4Ss1KuIZrs4i17pcNcKcRt-_6gJvk$. You are receiving this because you authored the thread.Message ID: @.***>

[jamesmkrieger]

Yes, exactly. Let me know if you have any problems

[TianmingQu]

Yes, exactly. Let me know if you have any problems

Hi James, I tried to use setEigens today with my eigenvectors and eigenvalues, the array of eigenvectors has a shape (number of atoms, 30) and the eigenvalues has a shape (30,).

When i do setEigens(eigenvectors,eigenvalues), it says the number of vectors and values do not match, is there anyway I can solve this problem?

[jamesmkrieger]

The vectors should be (3*n_atoms, n_modes). Probably that’s the problem. If you want only one coordinate per atom then you should set my_pca._is3d to False first

[TianmingQu]

How can I set this to false? By setting this as my_pca.setEigens(eigen_vectors, eigen_values, _is3d=True), like this?

Tianming

From: James Krieger @.> Date: Monday, March 6, 2023 at 4:19 PM To: prody/ProDy @.> Cc: Tianming Qu @.>, Author @.> Subject: Re: [prody/ProDy] About writing NMD file (Issue #1656)

The vectors should be (3*n_atoms, n_modes). Probably that’s the problem. If you want only one coordinate per atom then you should set my_pca._is3d to False first

— Reply to this email directly, view it on GitHubhttps://urldefense.com/v3/__https:/github.com/prody/ProDy/issues/1656*issuecomment-1457018216__;Iw!!Epnw_ITfSMW4!oRnWGNkgn5uqpubsbawNYgaGhISmuMrvCE6jSmT9Q6E7CffqZSnsk13xULL2YKjLn3rqtptJPGdKKlD-NUnk6n6rGYY$, or unsubscribehttps://urldefense.com/v3/__https:/github.com/notifications/unsubscribe-auth/AQXNBSSW4LLSNYIS3SGJW4TW2ZIHDANCNFSM6AAAAAAUVPYJFY__;!!Epnw_ITfSMW4!oRnWGNkgn5uqpubsbawNYgaGhISmuMrvCE6jSmT9Q6E7CffqZSnsk13xULL2YKjLn3rqtptJPGdKKlD-NUnkEIdr1GM$. You are receiving this because you authored the thread.Message ID: @.***>

[TianmingQu]

The vectors should be (3*n_atoms, n_modes). Probably that’s the problem. If you want only one coordinate per atom then you should set my_pca._is3d to False first

Actually I am facing a following problem:

I do pca to water density obtained from each frame and I wonder how I can visualized the fluctuation of water density PCs with Normal mode wizard. I now have the displacements of the water density along each PC and the eigenvalue and eigenvector of each PC.

Do you think there is anyway I can do this in Prody?

[jamesmkrieger]

You could maybe use two objects: one for eigenvectors and one for fluctuations.

The PCA one would be a standard PCA object and I guess the fluctuations would not be 3d.

You could maybe modify the fluctuations object before using setEigens or maybe use a GNM object instead instead of a 3d one.

[jamesmkrieger]

NMWiz shows GNM modes as colours so that may work for fluctuations.

[TianmingQu]

You could maybe use two objects: one for eigenvectors and one for fluctuations.

The PCA one would be a standard PCA object and I guess the fluctuations would not be 3d.

You could maybe modify the fluctuations object before using setEigens or maybe use a GNM object instead instead of a 3d one.

Could you be more specific on how to make these two objects?

[jamesmkrieger]

Try this:

obj1 = PCA() # is3d True so vectors1 is shape (n_atoms*3, n_modes) obj1.setEigens(vectors1, evalues)

obj2 = GNM() # is3d False so vectors2 is shape (n_atoms, n_modes) obj2.setEigens(vectors2, evalues)