LAUSR

Yang et al. reply — He and Bancroft have commented on shortcomings in the organization of the draft genome sequence of Brassica juncea, assembled using a hybrid strategy, and the characteristics of this genome evolution under crop selection1. B. juncea is a recent allopolyploid species that resulted from hybridization between Brassica rapa (A genome) and Brassica nigra (B genome) followed by chromosome doubling and evolution into vegetable and oilseed types for agricultural usage. We are aware of scaffold chimerism in this genome after checking the genomeordered graphical genotypes (GOGG). For better understanding of this allopolyploid genome, we have improved and updated a new version of the genome sequence of B. juncea, although the updated genome assembly does not impact the conclusions of homoeolog transcriptional behavior influencing selection in this allopolyploid B. juncea. He and Bancroft raised the possible problem of differentiating single nucleotide polymorphism (SNP) types for interhomeolog polymorphisms (IHP), interparalog polymorphisms (IPP) and allelic SNPs using 0.7-fold resequencing for genetic mapping in polyploids. In our analysis, we used 27-fold and 17-fold resequencing of two parent lines to differentiate IHP and IPP and identify allelic SNPs in parent lines. Only authentic allelic SNPs of two parents (aa × bb) were considered for further genetic map construction. Although there was low coverage of genome, 0.94-fold in an average of 100 individuals in F2 populations can accurately identify the integrated SNP genotypes (aa and bb) of parent lines after filtration. For each F2 individual, we consider the genotype of each block with sliding windows (windows, 100 SNPs; step, 50 SNPs). The genotype of the block (aa, bb, ab) is classified by the genotype rate and sequencing depth of all allelic SNPs (a,b,ab) in the block. Blocks having the same SNP genotypes are defined as a ‘Bin’2. This strategy can efficiently prevent the influence of IHPs and/or IPPs on the mapping outcome. Transcriptome sequencing can improve SNP identification in both gene sequences and gene expression in polyploids3,4. The genome-ordered graphical genotypes (GOGGs) platform developed by Bancroft can undoubtedly assess and improve genome assembly for complex polyploid genomes. We are aware that the pipeline of assembly we used initially was not sufficiently sensitive to certain scaffolds with few markers. We have specifically adjusted this pipeline to find candidate chimerisms for re-assembly of the genome. We split 426 scaffolds (total scaffold number 10,684) after checking the paired-end alignment of PacBio single-molecule and BioNano single-molecule reads around the breakpoints. We then constructed superscaffolds using BioNano molecules that retained superscaffolds of colinearity with the genetic map or with ancestor genomes if there were no genetic markers. Next, we constructed pseudochromosomes according to the genetic map. Confirming the genome organization with genome-ordered graphical genotypes of resequencing data supports an improved genome assembly (Fig. 1 and Supplementary Fig. 1). In combination with the GOGG platform and VHDH mapping population of B. juncea constructed by Bancroft, we anchored another 210 contig/ scaffolds into pseudochromosomes. According to the international convention of the Brassica community, we have adjusted the nomenclature of B subgenomes of B. juncea. After updating the genome, we compared conlinearity between subgenomes of B. juncea, its ancestor genome (B. rapa and B. nigra; Supplementary Fig. 2) and A subgenomes from B. rapa and B. napus (Supplementary Fig. 3), indicating an improved genome assembly. We updated the genome annotation, including genome validation with the CEG database gene model, transposable element (TE) and gene loss (Supplementary Tables 9, 14a,b, 16a,b, 17 and 23). We found that 23.4% of genes displayed differential homoeolog gene expression between subgenomes in all the same previously investigated samples (Supplementary Table 27b). These results support the conclusion of no global genome dominance, with differential homoeolog gene expression between subgenomes of allopolyploid B. juncea. The phenomenon of differential homoeolog gene expression between subgenomes appears to be conserved in allopolyploids5. With SNPs from the resequencing of B. juncea accessions based on the updated genome, we identified 890 selected genes between vegetable and oilseed subvarieties of B. juncea (Supplementary Table 27b). Among these, 26.2% displayed differential homoeolog gene expression between subgenomes of B. juncea. The re-analyses of transcriptional behavior and selection affirm the conclusion that differential homoeolog gene expression influences selection. ❐