Fig. 2
From: Cotton pan-genome retrieves the lost sequences and genes during domestication and selection
Multiple-scale variation for subgenomic divergence and GWAS on agronomic traits during cotton domestication. a Circos plot showing the SNP- and SV-based selection signals and QTLs during cotton domestication and improvement. The selection region was calculated in a 1-Mb sliding window with a step size of 200 kb. I–VIII, Circos plot from outer to inter tracks showing gene density (I), snpQTLs (II), cnvQTLs (III), the ratio of nucleotide diversity (π) based on SNPs between 256 landraces and 1364 improved accessions for domestication (IV), the ratio of nucleotide diversity (π) based on SNPs between 438 GhImpUSO accessions and 929 GhImpCHN accessions for improvement (V), the relative SV allele difference in the comparisons between landrace and improved accessions (VI), and between GhImpUSO and GhImpCHN (VII). The track (VIII) represents the domesticated homologous. Upper and lower panels (VI) represent deletion and duplication variation allele difference, respectively. The snpQTLs were identified using the meta-GWAS analysis of 890 cotton accessions. The outermost circle of the circos plot purple and yellow font shows pleiotropic snpQTLs (psnpQTLs) and pleiotropic cnvQTLs (pcnvQTLs), respectively. b–i Selective signals of copy number variations (CNVs) between the A (b) and D (f) subgenome during domestication. The horizontal gray dashed lines show the domestication signal threshold with the ratio of nucleotide diversity between wild/landrace and improved cotton accessions (πlandrace/πImproved > 200). c–e and g–i Six CNV-based GWAS hits that overlapped with domestication selection signals are shown for seed index (SI) (c), fiber length (FL) (d), boll weight (BW) (e), fiber uniformity (FU) (g), fiber elongation (FE) (h), and flowering date (FD) (i). The threshold of cnvQTL line was -log10 P = 4.4. The violin plot showed phenotypic variation with the lead CNV genotype. The numbers in the violin plot show the number of accessions for each copy. The significance difference was calculated with two-sided Wilcoxon rank-sum test (**P < 0.01, *P < 0.05)