|The upper uses two fields to identify a single record. The lower one uses a single field; they would allow different things.
I'd recommend to use the smallest set of fields required. Imagine you are the end-user, how do you differentiate between each record? What is it that makes a sample unique?
If a sample can only be identified by giving both a tbl_sample_id and a track_id, they'd need both be part of the key. If either can be used, then the smallest would become the key, and the other a unique index (the upper in the first case, lower in the second case).
Bastard Programmer from Hell
If you can't read my code, try converting it here[^]