We describe a global optimization approach for genome assembly where the steps of scaffolding, gap-filling, and scaffold extension are simultaneously solved in the framework of a common objective function. The approach is based on integer programming model for solving genome scaffolding as a problem of finding a long simple path in a specific graph that satisfies additional constraints encoding the insert-size information. The optimal solution of this problem allows one to obtain new kind of contigs that we call distance-based contig. We test the algorithm on a benchmark of chloroplasts and compare the quality of the results with recent scaffolders.