1.) How do I build a profile for multi-chain sequences?
You don't. These methods are designed to find single-chain templates, so
one approach would be to search for templates for each chain
individually, then combine all of the results into a multi-template
alignment.
2.) I mentioned this in the first part, but I'd like to use multiple
templates for this analysis. Do I used the PDB ID and each chain
associated with the TCR, when doing so?
Not sure what you're asking here. Each template would need a unique name
(align code) but you can call it whatever you want. Convention is to use
PDB ID + chain ID, but you're not required to do that.
Also how should I modify the
restraints argument between TCR chains?
What is the "restraints argument"?
3.) Accurate modeling of the CDR loops of the TCR (essentially the
active site) is important to my research, should I use loopmodel() or
other means for loop refinement?