How to predict binding sites from sequence data:
Get a collection of proteins of similar amino acid sequences and analogous biochemical function from your database.
Align these sequences amino acid by amino acid.
Check which regions of the protein are highly conserved in the course of evolution.
The binding site should be in one of the highly conserved regions.