Clustering trajectories and PCA data

[Former user]

Hi, i have simulated three protein-ligand complexes and performed PCA analysis on CA atom. But i had to do PCA three times for three trajectories. Is there any way to load multiple trajectories at a time for PCA analysis, and cluster their results together in order to have best comparative data/projection.

if it is possible please share code. i'm beginner of Bio3D and i'm following tutorial commands for understanding

[Xinqiu Yao]

Hi,

‌

It can be done of course. What you need is just to concatenate trajectories into one 'xyz' matrix. For example, xyz <- rbind(traj1, traj2, traj3). Then, do PCA with xyz.

‌

Let me know if you have further questions.

[Rimon Riju]

Hi ,

could you please explain a bit about it,

suppose i have two trajectories (pdb1, dcd1) and (pdb2,dcd2)

How do i cluster those trajectories for PCA?

you told me for xyz <-rbind (tarj1, traj2)

I did same but dcd is not clustered, however what about their two pdb structures?

for single pdb , the tutorial said xyz must be fitted with pdb,

so , what about my pdb fitting with?

‌

could you please help me by witting a code or sending an example ???

[Xinqiu Yao]

For clustering you need to call a function such as hclust() - it won’t automatically do it for you after PCA. Also, for trajectories with different pdb files, you need to align them first and do PCA on only aligned positions. Please take a look of bio3d tutorials (and possibly R related as well).

We are glad to answer questions directly related to bio3d but really do not have space to respond on requests for specific analysis tasks.