Genome-wide patterns of genetic variation in sweet and grain sorghum (Sorghum bicolor)

Abstract

Background: Sorghum (Sorghum bicolor) is globally produced as a source of food, feed, fiber and fuel. Grain and sweet sorghums differ in a number of important traits, including stem sugar and juice accumulation, plant height as well as grain and biomass production. The first whole genome sequence of a grain sorghum is available, but additional genome sequences are required to study genome-wide and intraspecific variation for dissecting the genetic basis of these important traits and for tailor-designed breeding of this important C4 crop.

Results: We resequenced two sweet and one grain sorghum inbred lines, and identified a set of nearly 1,500 genes differentiating sweet and grain sorghum. These genes fall into ten major metabolic pathways involved in sugar and starch metabolisms, lignin and coumarin biosynthesis, nucleic acid metabolism, stress responses and DNA damage repair. In addition, we uncovered 1,057,018 SNPs, 99,948 indels of 1 to 10 bp in length and 16,487 presence/absence variations as well as 17,111 copy number variations. The majority of the large-effect SNPs, indels and presence/absence variations resided in the genes containing leucine rich repeats, PPR repeats and disease resistance R genes possessing diverse biological functions or under diversifying selection, but were absent in genes that are essential for life.

Conclusions: This is a first report of the identification of genome-wide patterns of genetic variation in sorghum. High-density SNP and indel markers reported here will be a valuable resource for future gene-phenotype studies and the molecular breeding of this important crop and related species.

Figure 1

Genome-wide landscape of genetic variation in Sorghum bicolor. Gene density of chromosomes is visualized by line darkness; the more genes on a chromosome region, the darker the color. The purple and blue colors in the CNV ring represent gain and loss of copy number variation, respectively. For PAVs, the green color stands for the absence of variation, whereas pink for the presence of variation.

Figure 2

Number and distribution of non-synonymous and synonymous SNPs in different Pfam genes in the resequenced sorghum genomes. The Pfam gene families with 30 or more non-synonymous and synonymous SNPs were analyzed and are listed. The Pfam genes are arranged according to the percentages of non-synonymous and synonymous SNP sites. The top Pfam gene families have lower percentages of non-synonymous SNP sites, while the bottom ones have higher percentages of non-synonymous SNP sites. The numbers in the non-synonymous and synonymous horizontal bars show the absolute numbers of SNPs, whereas the numbers in the gene categories are the numbers of genes in each category. For each Pfam, the number of genes in the categories of bona fide genes, low-confidence genes, transposons and pseudogenes are also listed. Gene numbers that are lower than 5% of the total genes analyzed are not shown. The chi-square significance of the observed non-synonymous and synonymous SNP distributions for each Pfam group is shown: *P-value < 0.05; **P-value < 0.001.

Figure 3

Annotation and distribution of SNPs. (a) Pfam gene families significantly (P-value < 0.01) enriched or depleted with large-effect SNPs. Asterisks indicate Pfam families statistically significantly depleted of large-effect SNPs. (b) Statistics of different types of large-effect SNPs. (c) Statistics of synonymous and non-synonymous SNPs.

Figure 4

Distribution of 1- to 10-bp indels in the sorghum genome. (a) Number of indels of different length in the coding sequence (CDS) regions and the whole genome. (b) Number of genes that contain indels of different lengths. The figure reveals that 3-bp indels in CDS regions and genes that contain 3-bp indels are of the largest quantity. This implies that 3-bp indels cause the least negative effects on sorghum survival.

Figure 5

Chromosomal locations of genes differentiating sweet and grain sorghum. Genes with large-effect SNPs, indels and PAVs in the two sweet sorghum lines but devoid of these genetic variations in the grain sorghum were identified, considered as the sweet-associated genes, and mapped to the sorghum genome (see text for details). A 1-Mbp sliding window was used to define sweet-related regions on individual chromosomes, and only those windows containing more than three sweet-associated genes are shown. The overall gene distribution in the sorghum genome is shown by the grey bars as the background of every chromosome.