Hi Peter,
I personally do both individual PCA on the individual systems (so on the WT
and mutant in your example), and then combine the trajectories and perform
PCA on the combined set (find a common set of atoms for the WT and mutant
and trajin both sets into Cpptraj).
The first will tell me the independent systems' principal components, and
an rms inner product of the eigenvectors will tell me how similar they are
to other determined PCs.
The second will tell me what PCs the systems have in common.
A plot of PC1 vs. PC2 for two systems only makes sense to me if the PCs
they are describing are the same, so I would make this plot for the
combined PCA.
I use both in this publication, so you can read it and see what information
each of these approaches gets you:
https://www.tandfonline.com/doi/abs/10.1080/07391102.2023.2270749
The following tutorial gives an example of combined PCA that is described
in the second part of your question/my response.
https://amberhub.chpc.utah.edu/introduction-to-principal-component-analysis/
-Christina
On Thu, Dec 5, 2024 at 9:57 AM Ramdhan,Peter A via AMBER <amber.ambermd.org>
wrote:
> Hello,
>
> I have one question related to PCAs. The task I am currently working on is
> comparing an wild-type enzyme vs its mutant. I conducted 3x 500 ns
> replicates of each system for sampling purposes and concatenated the
> trajectories into a 1.5 microsecond trajectory for each system.
>
>
> 1.
> When conducting a PCA for these two, is it acceptable to perform the PCA
> on one and PCA on the other separately and then plot the PC1 vs PC2 from
> their respective pca.dat files on one graph? By comparing the overlap, I
> can see the space not explored by either system. Is this the right way to
> compare the two or do I need to concatenate both trajectories together and
> perform one PCA on a giant joint wt-mutant trajectory for analysis?
>
> Thanks!
>
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
--
-----------------------------------------------------------------
Christina Bergonzo
Research Chemist
Biomolecular Measurement Division, MML, NIST
-----------------------------------------------------------------
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Dec 05 2024 - 07:30:02 PST
