Fig. 3

Sequence gain and loss after polyploidization. a, b Unsupervised hierarchical clustering was used to identify diploid and tetraploid genome-specific kmers. The heatmap indicates the Z-scaled relative abundance of genome-specific kmers, and the color bar on the top axis indicates the kmer assigned to a given genome. c Schematic illustrating the increase or decrease in PAV sequences in cotton. d Distribution of intraspecific PAV sequences in the A_t subgenome. Some genes are shown. These sequences are absent in the A₂ reference genome but present in NRSs and tetraploid genome sequences. e The FBP and HDH variation in the A₂ pangenome and A_t subgenome. The pie chart represents the gene presence frequency in the 341 accessions. f, g A permutation test was used to assess the distance of lost sequences from genes in the A₂ and D₅ genomes. The green bars represent the observed lost sequence with gene distances, and the black bars represent the relationship between random sequences and gene distance. h GO enrichment of genes in lost sequences in A₂ and D₅