bioinfologics/sdg

Include circles when creating unitigs in the graph maker

Closed this issue · 2 comments

Cases where a canonical kmer has a canonical neighbour that is itself, will get dropped from the graph currently with the new dbg building algorithm. E.g ATA has a fw AND bw neighbour TAT (canonically ATA), and so the unitig ATAT would be dropped from the graph. This is an example of the more general circle problem: at the end of the unitig creation, any kmers that are not used will be in circles: We can add those circles as unitigs by choosing a random break point.

when the circles are simplest (i.e ATATATATA.... etc) a single kmer is created but it is not connected

solved on 4b64d1c