PCA for multiple PDB file by using MDanalysis

[JUNJIE ZHANG]

Hi experts,

I am wondering that how can i do pca analysis for multiple PDBs.Assume that currently I have 1000 pdbs, and i want do dimensionality reduction for those files?

Bests,

Junjie

[Oliver Beckstein]

Hi Junjie,

Welcome to the MDAnalysis list!

On Jun 7, 2023, at 2:31 AM, JUNJIE ZHANG <junji...@gmail.com> wrote:

Hi experts,

I am wondering that how can i do pca analysis for multiple PDBs.Assume that currently I have 1000 pdbs, and i want do dimensionality reduction for those files?

If they have all the same structure then you can treat them as a “trajectory” and just use the standard MDA PCA.

u = mda.Universe([structure1.pdb, structure2.pdb, …])   

ca = u.select_atoms(“name CA”)

# then do PCA , see https://userguide.mdanalysis.org/stable/examples/analysis/reduced_dimensions/pca.html

If the structures are different then you have to do more work because MDAnalysis works on trajectories. You could create a trajectory only containing the common subselection from your structures. We’re happy to help when you have specific questions on how to go about doing that.

Alternatively, you could also look into tools outside of MDAnalysis such as ProDy http://prody.csb.pitt.edu/ which might do what you need — I don’t have personal experience with it.

Best,

Oliver

Bests,

Junjie

--
You received this message because you are subscribed to the Google Groups "MDnalysis discussion" group.
To unsubscribe from this group and stop receiving emails from it, send an email to mdnalysis-discus...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/mdnalysis-discussion/6e49c2d4-6e63-44ec-b1fa-df651442385an%40googlegroups.com.

--

Oliver Beckstein (he/his/him)

email: orbe...@mdanalysis.org
twitter: @orbeckst

GitHub: @orbeckst

MDAnalysis – a NumFOCUS fiscally sponsored project

https://www.mdanalysis.org/