PCA from nolecular dynamics and structure comparison

[Former user]

Dear all, I've been struggling with the following issue: I have performed a PCA of a molecular dynamics trajectory of my protein (a GPCR) and, separately, a PCA of homologous GPCR's structures. Now, what I want to do is to somehow concatenate these homologous structures to my own protein+trajectory to do a PCA of this whole system. What I aim to observe is how well my simulation was able to scan through the phase space of possible conformations based on the positions of the structural homologs in the pc plot. I have been trying to bring these two pdbs variables together using the tutorials as references but to no avail. Thank you for your time and attention!

[Xinqiu Yao]

Hi,

‌

It probably sounds a bit abstract without any specific example, but what I did is:

1. Do PCA for the crystal pdbs using all “non-gap” positions (or “core” positions depending on which strategy you wish to use).
2. Align the sequence of the PDB used for MD to the existing crystal pdbs. At the same time, find the C-alpha atom indices of the PDB that match the “non-gap” positions used for PCA. You can use pdb2aln.ind()to do it.
3. Then, superimpose (fit) all MD structures to the first structure in pdbs using the indices obtained from step 2.
4. Project the superimposed trajectory coordinates to the PC subspace using project.pca(), along with the C-alpha indices from step 2.

‌

Type ?pdb2aln.ind and check the Examples section. It does some similar steps.

‌

Hope it will help.

‌