Hierarchical decomposition improvements #164
Labels
hierarchical decomposition
it took me a year to implement this so it had better be useful .___.
lowpriorityfeature
It works pretty well now, but as I've been working with the interface over the past few days I see a few things which should be fixed.
Disallow frayed ropes from being used in other patterns (including in other frayed ropes): IMO the fact that they have multiple start/end points makes them ambiguous.
If a chain has already been identified and it's a candidate as the start/end node of a bubble, this should be allowed. It isn't currently. Use case: component 75 of the first biofilm graph, contig_0000016197 should form a bubble to contig_0000011283 but since both are in chains this doesn't work.
Allow cyclic chains to be boundary nodes of bubbles (same as with chains).
Likely related to the above: detect cyclic chains of bubbles. This is a real pattern present in some of the components of the test Velvet E. coli graph, and currently we're detecting 2/3 of the bubbles. Ideally we'd catch all three bubbles and wrap them in a cyclic chain. (What we have now is still, imo, better than what we had previously.)
Old:
Current:
I am not sure 100% why this is ending up like this; may be a bug in pattern decomp somewhere.
Probably more things I'm not thinking of / haven't run into yet
Honestly, this already looks pretty nice, so these fixes are low priority since they're pretty subjective
The text was updated successfully, but these errors were encountered: