I wanted to explain the process of implementing PCA (Principal Component Analysis) on your MD (molecular dynamics) trajectories. Here I do it on NAMD generated DCD files. The example files are taken from here. This tutorial can be considered as a Matlab version of the bio3d R package tutorial example from Grant Lab. The steps are as below :
Access the files using readdcd command. This is not a Matlab provided command. You can download it from here.
The hivp.dcd and hivp.pdb files in the Grant Lab example can be downloaded from here.
FIT THE DCD files. THIS IS REALLY IMPORTANT. I DID IT USING tcl scripting in VMD.
Then you may use PRINCOMP command as below to do PCA. VOILA!!
PRINCOMP requires Statistics toolbox, so I will explain a method with no princomp in future, when time permits. :D
clear,close,clc
xyz=readdcd('/your/location/fitted_hivp.dcd',1:198);
% FITTING DCD IS REALLY REALLY IMPORTANT !!!!!!!!!
[pc, score, latent, tsquare] = princomp(xyz(2:end,:));
latent(1)/sum(latent)% PC1 percentage
latent(2)/sum(latent)% PC2 percentage
latent(3)/sum(latent)% PC3 percentage
plot(score(:,1),score(:,2),'.')
xlabel('PC1')
ylabel('PC2')
