Fig. 2: A schematic diagram of graph pruning in KSNP (k = 2).

The blue and orange nodes represent 2-mers from haplotypes H0 and H1, respectively, where the two numbers in a node indicate the corresponding position indices of the two consecutive SNPs. The solid and dashed arrows represent edges that are retained and removed in each processing step, respectively. The numbers on the arrows denote the depth of the edges. This graph is horizontally symmetrical, indicating the two complementary haplotypes in a diploid genome. a The competitive edges (dashed arrows) at the same SNP position are removed since they have very shallow depth in comparison with the dominate path. b The short tip is deleted if the length of the long path exceeds three times that of the short path. For example, Tip1 and Tip2 are deleted given Long path 1 and Long path 2, respectively. c For a bubble, KSNP attempts to find an optimal path from the source node to the sink node under the supervision of MEC scores. For example, KSNP calculates the MEC scores of the three possible haplotypes in the illustrated nested bubble using the reads involved in the bubble (not displayed), and the nodes and edges that are not located on the best path are removed. For a long branch like the one from node [10,11] to node [13,14], by restoring a previously removed edge between node [12,13] and node [13,14], the branch can be completed into a bubble and then resolved as a bubble. d The haplotype blocks are generated by walking out the paths in the pruned graph.