Scalable variant detection in pangenome models

We have implemented a two-step scalable approach to detect variants: first we construct a graph pangenome from a graphical fragment assembly (GFA) file that stores the fragments, where each fragment corresponds to a vertex of the graph, then we analyze the graph to detect all variants. We have tested our approach on a SARS-CoV-2 dataset with over 7800 fragments and on a dataset that contains all alternative sequences of the highly polymorphic human leukocyte antigen (HLA) complex.