How do gap penalties change a protein alignment?

Why

An affine gap model typically charges an opening cost plus an incremental extension cost for each additional gap position. With a large opening penalty and small extension penalty, the algorithm often prefers one contiguous insertion over many scattered gaps. This can be appropriate for a loop insertion but can misplace residues when repeats or low-complexity segments occur.

For protein alignment, use a substitution matrix such as BLOSUM together with a defensible gap model, then assess conserved catalytic residues and structural boundaries. A changed alignment is not merely cosmetic: it can alter inferred mutation positions, domain boundaries and homology claims. Inspect alternatives and known motifs, especially when gaps cluster next to active sites.

There is no single universal penalty pair for every evolutionary distance or sequence type.