per-residue eigenvalues in PCA?

[Alireza Tafazzol]

I have 2 questions:

1. In the PCA (pca.xyz) of a loaded trajectory, is there a way to get the contribution of each residue (atom set) in the first eigenvalue (PC1)? (i.e, per-residue or per-atom eigenvalues of PC1?)
2. When making a projected trajectory “mktrj.pca” on the PC1, is there a way to have values for each residue?
   I was trying to open the coordinate file of saved PDB file and calculate the thickness/distance between any two corresponding atoms (x,y,z) in the produced models. This way the summation of all thicknesses should be equal to the 1st eigenvalue (PC1), but it is somehow different. Should the magnitude of this projected trajectory be equal to the first eigenvalue ??

‌

[Xinqiu Yao]

Hi,

‌

1. The atom-wise loadings should be what you are looking for. Please check the help document ?pca for more detail
2. I am not sure what you were measuring here. The magnitude of the movement is scaled by the square root of the eigenvalue, and so I guess you might need to sum the squares to get equal to the eigenvalue.

If what you need is the contribution of each residue to the eigenvector, I think the loadings should be the answer.

‌

Hope it helps.

‌