By examining the structure and referring to the publication about the structure, you can see that the statin binding site is located at interfaces between two monomers. Therefore, you need to retain at least two of the chains that form a given binding site for the inhibitor.

The presence of many chains in a PDB does not necessarily define that the protein is multimeric. You only need one chain for molecular docking. Even in homomeric proteins, a chain should be selected. For this, emphasis should be given to binding site residues. A chain that possesses an intact structure and all the binding site residues should be selected for better docking results.

It all depends upon the fact that where the active site residues lie in the structure. This knowledge could be achieved by thorough literature survey. After finding out your target residues, yes you can select that particular chain(s)

Thank you Tehreem Tahir Bimal Kumar Raut Rudy J Richardson . I am planning to perform blind docking to see if there are other interesting binding sites where my ligands can attach to. Any comment regarding my method?

I've used the same approach in my research. It gives you insight about mode of inhibition (competitive or un competitive)

Jack Dela Cruz : Yes, blind docking can be used to locate potential alternative binding sites for ligands. My colleagues and I have used this approach to find potential allosteric sites in the dimeric Mpro protease in the SARS-CoV-2 virus. However, blind docking results should be interpreted with caution, as there will be non-specific binding to loci that are neither active sites nor allosteric sites. I would also like to emphasize that while active sites of proteins are often located within a single protein chain in dimeric or multimeric proteins, there are cases for which the active site is formed at protein-protein interfaces. Examples of the latter include your protein of interest, HMG-CoA reductase, as well as human transthyretin and human ornithine decarboxylase.