Is it possible to perform an automatic search of a multi-chain PDB
template with a known protein sequence that is to be modeled (say
100-200 residues in length) and then extract the chain ID of the segment
of the PDB template that best matches with the query sequence.
If I understand you correctly, this is straightforward. If you're
scanning against our PDB95 database, for example, all the templates are
single chain, and the chain ID is simply the last character in the align
code.
If you're trying to align against a multi-chain template (presumably a
local alignment) you could extract the chain ID trivially from the
alignment object.