An integrative approach to genomic introgression mapping

Abstract

Near-isogenic lines (NILs) are valuable genetic resources for many crop species, including soybean (Glycine max). The development of new molecular platforms promises to accelerate the mapping of genetic introgressions in these materials. Here, we compare some existing and emerging methodologies for genetic introgression mapping: single-feature polymorphism analysis, Illumina GoldenGate single nucleotide polymorphism (SNP) genotyping, and de novo SNP discovery via RNA-Seq analysis of next-generation sequence data. We used these methods to map the introgressed regions in an iron-inefficient soybean NIL and found that the three mapping approaches are complementary when utilized in combination. The comparative RNA-Seq approach offers several additional advantages, including the greatest mapping resolution, marker depth, and de novo marker utility for downstream fine-mapping analysis. We applied the comparative RNA-Seq method to map genetic introgressions in an additional pair of NILs exhibiting differential seed protein content. Furthermore, we attempted to optimize the comparative RNA-Seq approach by assessing the impact of sequence depth, SNP identification methodology, and post hoc analyses on SNP discovery rates. We conclude that the comparative RNA-Seq approach can be optimized with sufficient sampling and by utilizing a post hoc correction accounting for gene density variation that controls for false discoveries.

Figure 1.

Chromosomal positions of Affymetrix SFPs and GoldenGate SNPs identified between Clark and IsoClark. Chromosomes are labeled at the top according to number, and centromere positions are shown as white circles. Red lines indicate the physical map positions of Affymetrix SFPs, and blue lines indicate the physical map positions of GoldenGate SNPs. Genomic regions coincident for both SFPs and SNPs are indicated with yellow boxes, and genomic regions exhibiting only GoldenGate SNPs are indicated with white boxes.

Figure 2.

Significant intervals of SNP clustering between the Clark and IsoClark lines were found on six chromosomes, 3, 4, 5, 8, 13, and 16, as determined from the bootstrap method. Chromosomes are labeled on the left according to number, and centromere positions are shown as white circles. Vertical boxes indicate 500,000-nucleotide intervals. The number of SNPs found in each interval is indicated above the interval. A, Clustering of SNPs obtained from the 10-d root data using method 1 on the single-library comparison. B, Clustering of SNPs obtained from the 10-d root data using method 2 on the single-library comparison. C, Clustering of SNPs obtained from the 19-d root and leaf data using method 1 on the four-library comparison. D, Clustering of SNPs obtained from the 19-d root and leaf data using method 2 on the four-library comparison.

Figure 3.

Significant clusters of SNPs for the seed protein lines were found on three chromosomes, 16, 18, and 20, as determined from the bootstrap method. Chromosomes are labeled on the left according to number, and centromere positions are shown as white circles. Vertical boxes indicate 500,000-nucleotide intervals. SNPs were identified via method 2. The number of SNPs found in each interval is indicated above the interval. SNPs were clustered from the seed protein RNA-Seq data that contained 14 libraries for each NIL.