I am trying to perform a PCA analysis using side-chain heavy atoms, instead of the conventional CA-based PCA. The question is in that case, do I need to fit the trajectory on a side chain reference or I need to do the fitting again on the backbone. My protein is rigid but interface residues adopt different rotameric states upon binding with different partners and my goal is to quantify these states through PCA to remove the noise associated with the MD trajectory. It will be great if someone can suggest some tools as the tools I am using (bio3d, Wordom) somehow gave different results...Thanks
You can calculate the RMSD trajectory (of backbone and side-chain heavy atoms) using a) backbone and b) side-chain heavy atoms for fitting. Stability of rmsd trajectory will indicate if least square method (for removing over all rotation and translation) of fitting for each set of atom selection converges or not.
Dear Dr. Hiqmet..thanks for the answer..but why shall I test fitting convergence on the subset of atoms I am not going to PCA them..my understanding is that you do fitting to get rid of the rotational/translational movements so that only unique motion is taken into account in covariance matrix diagonalization. Even if backbone convergence happened to be better, shall I use this backbone fitted trajectory to calculate the PCAs for sidechain atoms? Regards..Marawan
Because the CA or Backbone atoms are not involved in complex motion during the course of MD simulations (such as, tumbling motion). On the other hand, the side chain atoms have a complex rotation motion, when you apply the algorithm to remove this overall tumbling motion in side chain heavy atoms, then the algorithm itself (least square fit) may not converge. So, reaching the plateau (I do not necessary mean the value of rmsd at the plateau) in the rmsd trajectory could tell something about the convergence.
hmmm...I got your point and it is indeed convincing..do you think i should do the same if the deep problem i want to study, is how a single protein interacts with different partners, other proteins, small molecules, etc..no major global changes are observed, only the side chain rotamers are different in different complexes...By the way I had a look on your profile and came a cross a very interesting paper you published "Extracting the Causality of..." which talks about how to use information theory to decide, in correlation analysis, that it is a given residue that controls the other one...I think I will consider doing something similar also, as i was always thinking that simply saying that a pair of residues has a correlated motion is not enough to study something deep..did you implement this method in any software..Thanks in advance..Regards...Marawan
- Badges
- Science topic
- Similar topics
- Chemistry
- Biochemistry
- Biochemical Techniques
- Blotting Techniques
- Immunoblotting
- Dyes