PCA for tetramer

[Former user]

Hi There,

I have a number of trajectories for a tetramer protein and would like to do a PCA for the whole system, instead of doing it individually for four segments. Is there a way to do it using bio3d?

Thanks, Subha

[Xinqiu Yao]

Could you describe a bit more details about your problem since I don't see why you cannot do it with bio3d as for single subunit systems?

[skal24]

Hi Yao,

I am aiming to do a pca analysis for a homo tetramer (ion channel) protein and mostly interested in sampling the variations in the pore domain.

PCA analysis for each chain (or segment) is doable and also useful when I generate tetramers for each of the sampled (dominant) conformations (from each chain). This will provide me various symmetrical conformations of the channel. But, I would also like to see the variations in the chains with respect to one another, so that the asymmetry in the channel architecture is also captured.

I am wondering, if there is a way I can make Bio3d understand the segment breaks and do a PCA analysis combining all of the four chains(or segments)?

Thanks, Subha

[Xinqiu Yao]

If you have trajectories for the tetramer, why not just use them for pca? It will take care of chains and capture variations of the entire tetramer. Or, is my understanding wrong?

[skal24]

No, you are right. I did not try performing PC for the tetramer yet. I wasn't sure if bio3d could differentiate the chains from segment name and write pdb files in a similar way after clustering. That would be great, if it does that.

And, is there a way to know the contributions of each segment to the overall PC (i.e. from PC of tetramer)?

Thanks, Subha

[Barry Grant]

It works just as it should for single chains - it is up to you to provide the input you want to use. Please try it first and then let us know if you have more specific questions...

[skal24]

Since all four chains has the same residue numbers, writing out the residue contributions to the PCs doesn't give the expected results.

For example, Residue contributions to PC1

pro_au1<-tapply(pc_pro$au[,1],pro$atom$resno,max) pro_au1<-as.numeric(pro_au1) plot.bio3d(pro_au1,resno=pro)

gives me a warning, which is probably because of the number of residues. The tetramer has 4 times the no.of residues, whereas the plot is mapping the contributions of all four segments in one.

I get a plot with x axis numbering from 1 .. n instead of the resno labelling.

Warning message: In plotb3(...) : Length of input 'resno' does not equal the length of input 'x'; Ignoring 'resno'

How can I get it to plot the contributions for residues from all four segments?

Thanks, Subha

[Xinqiu Yao]

You can temporarily renumber your pdb using e.g. clean.pdb() or convert.pdb(), and then use the modified pdb for the calculation.

The pdb in calling plot.bio3d() must be the same tetramer, otherwise gives the Length of input 'resno' does not equal the length of input 'x' warning...