pyani classifyImplement graph theory classifier (cliques)

[widdowquinn]

Summary:

The pyani classify command can be sped up using the following method:

1. Create a complete graph from all comparisons
2. Apply initial threshold (e.g. 50% coverage) for removing edges
3. Identify graph components and, for each new component of size $n$:

• identify minimum number of node edges (if not $n - 1$, then this cannot be a clique)
• identify maximum number of node edges (if not $n - 1$ then this cannot be a clique)
• if component could be a clique, test for this property
  • if component is a clique, record it
• when all components are tested, apply the next threshold (e.g. delete edge(s) with lowest score) and repeat [END when all nodes are singletons]

This reduces the number of costly clique calculations