Consider the following protonation states of 3-amino 5-methyl indazol, in smiles format:
If I paste each of these compound variants one at a time into MarvinSketch (as SMILES), and then use the tautomer tool to ask for the dominator tautomer at a pH of 7.4, I get two very different answers (please refer to the enclosed picture). The dominate tautomer from the latter protonation state produces an imino form, which, by casual inspection, looks somewhat unlikely.
Naively, I would have expected both protonation states to be roughly equally likely.
Depending on how the two compounds are created (pasted from SMILES or drawn from scratch), different results can be obtained, suggesting some internal inconsistency.
I am currently using MarvinSketch 5.3.3.