I started using pairwise2.align function from Biopyton with the one_alignment set to True.
I was wondering if that single alignment returned is the best one. With Smith-Waterman it should be but I'm not sure if that's what's been implemented.
There may be multiple pairwise alignments that have the same optimal score with respect to gap penalty and match scoring. The function just returns the first it finds during the traceback of the dynamic programming matrix rather than returning all of them.
From the following link, several paragraphs down, you will find this quote:
"Our variable alignments now contains a list of alignments (at least one) which have the same optimal score for the given conditions."