i have performed MD simulation in GROMACS of the apo protein and another one with the same protein but with a ligand and performed PCA analysis as follows:
1- calculate the covariance matrix of the Calpha only of the apo trajectory alone using this command:
gmx covar -mwa -s ref.pdb -f step5_production_noPBC_Calpha.xtc -v eigenvectors_all_noref.trr -o eigenvals_all_noref.xvg
2- performed 2D projection of the first two eigenvectors from the apo trajectory using this command:
gmx anaeig -s ref.pdb -f step5_production_noPBC_Calpha.xtc -2d 2dproj_12.xvg -first 1 -last 2 -v eigenvectors_all_noref.trr -eig eigenvals_all_noref.xvg
3- to see the motion captured in the first 3 eigenvectors of the apo trajectory, I used this command:
gmx anaeig -extr extreme.pdb -first 1 -last 3 -s ref.pdb -f step5_production_noPBC_Calpha.xtc -nframes 30 -v eigenvectors_all_noref.trr -eig eigenvals_all_noref.xvg
the same three commands were done for the complex trajectory with changing the .xtc and using the eigenvectors and eigenvalues produced when i used the gmx covar on the trajectory of the complex (Trial 1).
then i did just the last two commands on the complex trajectory but this time i used the eigenvectors and eigenvalues produced from using gmx covar on the apo protein trajectory (Trial 2)
which way i can use to compare the PCA results of the two trajectories?
Thanks
When performing Principal Component Analysis (PCA) on two trajectories of a protein alone and a protein-ligand complex, you typically want to use the eigenvectors corresponding to the largest eigenvalues. These eigenvectors are associated with the most significant variations in the protein conformations and represent the dominant modes of motion in the system.
The steps to perform PCA on two trajectories are as follows:
It is worth noting that the choice of eigenvectors will depend on the specific requirements of your study and the size of your datasets. Additionally, PCA is just one of many tools that can be used to analyze molecular dynamics trajectories, and you may need to use other techniques, such as clustering or free energy calculations, to gain a complete understanding of the protein's motion in the presence and absence of the ligand.
My suggestion would be to concatenate all the trajectories (apo and ligand-bound) and on that meta trajectory, calculate the covariance matrix of the coordinates and diagonalize it to obtain the desired eigenvectors. That way, you could immediately see the ligand's effect on the first two eigenvectors.
Further reading:
Palma, J., & Pierdominici‐Sottile, G. (2023). On the Uses of PCA to Characterise Molecular Dynamics Simulations of Biological Macromolecules: Basics and Tips for an Effective Use. ChemPhysChem, 24(2), e202200491.
Cossio-Pérez, R., Palma, J., & Pierdominici-Sottile, G. (2017). Consistent principal component modes from molecular dynamics simulations of proteins. Journal of Chemical Information and Modeling, 57(4), 826-834.
Edward Francisco Méndez Otálvaro Satyendra Singh thank you for your help. i searched and found that they do as Edward Francisco Méndez Otálvaro said
- Badges
- Science topic