In #906, @Vidishab18 showed some test cases using AlignmentTools.toConciseAlignmentString can produce different results depending on HashMap iteration order.

The method takes an alignment map (a way of representing a pairwise alignment as a map between residue indices) and outputs a string representation. The representation is not unique. For instance "1>2>3>2 4>3" and "4>3 1>2>3>2" are the same alignment. Likewise, "1>2>1" and "2>1>2" are the same alignment. Outputs potentially hinge on the enumeration order of the underlying map.

Possible solutions:

1. Create a canonical form (e.g. sorted some way). I'm not sure whether this is feasible; general graph canonization is computationally hard, but perhaps the special structure of alignment graphs gives an obvious order
2. Use a SortedMap (TreeMap) internally. This should make it deterministic, although perhaps not in a simple form
3. Merge Fixing flaky test in AlignmentToolsTest.java #906, which apparently fixes the tests even though it leaves the function itself unchanged.

The method was intended as a concise human-readable output of the map, so the order doesn't really matter. Option 2 appears to be the simplest solution. Performance on this method is not critical.