How to explain different PCA results by projecting onto reference eigenvector

GROMACS version: 2021.1
GROMACS modification: No

Hello community,

I want to study the single pint mutation (or double mutations) effect on the structural conformation.
Suppose the protein name is prot_A. I obtained the wild type pdb structure from rcsb pdb (named WT.pdb) and I did
the single point mutation from its sequence and build the homology model from swiss-model (named MUT.pdb).

I did 300-ns Gromacs simulation by following the tutorial (Lysozyme in Water; Lysozyme in Water).

Finally, I performed PCA study in order to compare the 2D PCA plot.

Learning from this forum (@alevilla), I need project both WT and MUT proteins production trajectory onto the same
eigenvector (only backbone with same number of atoms. Either WT as reference eigenvector or MUT as reference eigenvector) for comparison.

I did both ways as below. For the 1st experiment, I used the WT as the reference structure and reference eigenvector,
and in the 2nd experiment, I used the MUT as the reference structure and reference eigenvector.

After obtaining corresponding xvg files, I make the 2D PCA plot and calculate the approximate areas (convex hull).

################################

WT as the reference

################################

Extract reference backbone structure from WT

echo 4 | gmx trjconv -s WT.tpr -f WT.xtc -o WT_bb_ref.pdb -dump 0

Extract reference backbone eigenvector

echo 4 | gmx trjconv -s WT.tpr -f WT.xtc -o WT_bb.xtc
echo 0 0 | gmx covar -s WT_bb_ref.pdb -f WT_bb.xtc -o WT_eigenval_bb.xvg -v WT_eigenvec_bb.trr

Project to both reference structure and eigenvector

echo 0 0 | gmx anaeig -s WT_bb_ref.pdb -f WT_bb.xtc -v WT_eigenvec_bb.trr -2d WT_pca_2d.xvg -first 1 -last 2
echo 0 0 | gmx anaeig -s WT_bb_ref.pdb -f MUT_bb.xtc -v WT_eigenvec_bb.trr -2d MUT_pca_2d.xvg -first 1 -last 2

################################

MUT as the reference

################################

Extract reference backbone structure from MUT

echo 4 | gmx trjconv -s MUT.tpr -f MUT.xtc -o MUT_bb_ref.pdb -dump 0

Extract reference backbone eigenvector

echo 4 | gmx trjconv -s MUT.tpr -f MUT.xtc -o MUT_bb.xtc
echo 0 0 | gmx covar -s MUT_bb_ref.pdb -f MUT_bb.xtc -o MUT_eigenval_bb.xvg -v MUT_eigenvec_bb.trr

Project to both reference structure and eigenvector

echo 0 0 | gmx anaeig -s MUT_bb_ref.pdb -f WT_bb.xtc -v MUT_eigenvec_bb.trr -2d WT_pca_2d.xvg -first 1 -last 2
echo 0 0 | gmx anaeig -s MUT_bb_ref.pdb -f MUT_bb.xtc -v MUT_eigenvec_bb.trr -2d MUT_pca_2d.xvg -first 1 -last 2

PCA plot

My questions are:
(1) In the 1st row of the figure (WT as the reference structure and eigenvector), I found that MUT protein showed the
smaller area by comparing WT protein. Can I get a conclusion that MUT protein limited the protein flexibility comparing with WT?
Or can I say that the mutation on WT protein made the protein more rigid?
(2) In the 2nd row of the figure (MUT as the reference structure and eigenvector), I found the conclusion is just opposite where
it seems that WT gave the smaller area by comparing the MUT. Can you help me figure out what is the possible reason for this different
conclusion?
(3) Can I directly compare the PCA plot area in order to comparing the MD trajectories for WT and MUT proteins?
(4) Is PCA a good method to study the effect of mutation on protein structural conformation? It will be helpful if giving me some suggestions
on what metrics/methods/techniques can be used to study the mutation effect on structural conformation.

Sorry for these questions but I am just a new GMX user.

Thanks.

Hello,
I want to do something similar with my system and was just wondering if you figured out how to do this? I would be grateful if you could share with me how this is done.

Best wishes,
Pawel