Problems in graph contraction for 5xnl

[josemduarte]

For 5xnl, during graph contraction, some homomeric edges are chosen to be contracted:

 WARN  eppic.assembly.GraphContractor - This looks like a homomeric interface! We should not be contracting it, something is wrong!

Only heteromeric edges should be contracted, so this must be an error.

Another example: 2ian

[josemduarte]

In fact, for 2ian the assembly enumeration with contraction gives currently the wrong result for the second assembly id=2. That's because the edges chosen for contraction must be isomorphic throughout the graph, see solution now implemented in 73fac5c. 5xnl still has the problem mentioned above even after that commit.

[josemduarte]

The issue is less severe now after the above commits. However there are still problems with the assembly enumeration for 5xnl and for a bunch of other large heteromeric entries.

This is a list of entries that potentially have similar problems (as of 7th Oct 2017):

5u0a
5afu
1fnt
5iv7
5iv5
5mqf
3jb9
5adx
5voz
5vox
5voy
5fij
5fik
5fil
4d6t
5xjc
2gtl
4v56
4v54
4v5f
4v89
3j9v
3j9u
3j9t
5kgf
4h1l
4b2q
3jc9
3jc8
1z7q
5nw4
5ari
5are
5ara
5arh
5wfe
5gmk
5gm6
5nl0