Fig. 1

Scheme diagram of PSVCP pipeline. a Construction of linearized pangenome. Genomes used for pangenome construction are iteratively aligned between genomes and the starting reference to identify novel segments, then integrate these sequences into the reference genome to construct a pangenome. b PAV genotyping in the pangenome for single accession. PAVs are detected by mapping sequencing reads of each accession to the pangenome and calculating mapping coverage using a 20-bp window. Two adjacent 20-bp regions are merged if they show the same presence or absence pattern. c Population-wide PAV genotyping. The breakpoint positions of each accession are recorded, and a complete set of all breakpoint positions (the breakpoints’ union) are generated. Every segment with two adjacent breakpoints from the breakpoints’ union is defined as a new PAV region, named by the adjacent left breakpoint position. PAVs in each accession are re-called using the integrated PAV genotypes. All accessions’ new PAV genotypes are combined by row, generating a PAV genotype matrix with accession names as row labels and the adjacent left breakpoint position as column labels. The PAV genotype matrix was further filtered by minor allele frequency (MAF) >0.05. d Translocation genotyping by PSVCP. The novel insertions are aligned against the pangenome using BLAST+. The novel insertions matched to 2 or more sequences on the pangenome are identified as potential translocations. The presence or absence of translocations is determined by looking at the read mapping around its breakpoints, spanning a 39-bp region with a conjunction point at the center. e Inversions genotyping by PSVCP. The inversions are identified by comparing each genome used for pangenome construction with the pangenome using assemblytics. We further genotyped each potential inversion on the pangenome from 413 accessions by examining the mapping coverage of the 39-bp region with breakpoints in the center