Illumina short-read assemblies for microbial genomes are normally suffered from fragmentation due to repetitive elements that require long reads or close reference genomes to resolve. While long-read sequencing data is not always available, reference-based methods are usually subject to the bias problem which can be severe for highly diversified strains.To tackle this issue, we use the pangenome graph built from a large set of multiple related genomes to guide the scaffolding process of a bacterial strain genome assembly. We propose Pasa, a graph-based algorithm framework that can utilize the pangenome graph and the assembly graph information to improve scaffolding quality.