Fig. 5

Scaffolding path reconciliation using constituent gene models. Each contig is denoted by a letter in a circle. The blue and green boxes represent the gene models at the 5′ and 3′ ends of a contig. Joining-pairs connecting two contigs are shown in purple, and orientation is indicated by arrows. Contigs are reverse-complemented as needed. a The scaffolding path obtained by following highest-weighted edges. Examining the gene model between each pair of the contigs in the path tells us that the extension from A to D violates the requirement for zero intervening gene models between two contigs. Therefore, the reconciled path contains only two contigs, rather than four. b The current best path is not the optimal one because it incorporates only a subset of all vertices. AGOUTI therefore picks another vertex and re-walks the subgraph. After reconciliation, the new path becomes the best path as it has more vertices than the previous one. c Similarly, AGOUTI next starts from node D and establishes a new path. The reconciled path contains all four vertices in the subgraph, and therefore AGOUTI uses it as the optimal one (edges shown in red) and stops checking other possible paths