Exome genotyping, linkage disequilibrium and population structure in loblolly pine (Pinus taeda L.)

Abstract

Background: Loblolly pine (Pinus taeda L.) is one of the most widely planted and commercially important forest tree species in the USA and worldwide, and is an object of intense genomic research. However, whole genome resequencing in loblolly pine is hampered by its large size and complexity and a lack of a good reference. As a valid and more feasible alternative, entire exome sequencing was hence employed to identify the gene-associated single nucleotide polymorphisms (SNPs) and to genotype the sampled trees.

Results: The exons were captured in the ADEPT2 association mapping population of 375 clonally-propagated loblolly pine trees using NimbleGen oligonucleotide hybridization probes, and then exome-enriched genomic DNA fragments were sequenced using the Illumina HiSeq 2500 platform. Oligonucleotide probes were designed based on 199,723 exons (≈49 Mbp) partitioned from the loblolly pine reference genome (PineRefSeq v. 1.01). The probes covered 90.2 % of the target regions. Capture efficiency was high; on average, 67 % of the sequence reads generated for each tree could be mapped to the capture target regions, and more than 70 % of the captured target bases had at least 10X sequencing depth per tree. A total of 972,720 high quality SNPs were identified after filtering. Among them, 53 % were located in coding regions (CDS), 5 % in 5' or 3' untranslated regions (UTRs) and 42 % in non-target and non-coding regions, such as introns and adjacent intergenic regions collaterally captured. We found that linkage disequilibrium (LD) decayed very rapidly, with the correlation coefficient (r (2)) between pairs of SNPs linked within single scaffolds decaying to half maximum (r (2) = 0.22) within 55 bp, to r (2) = 0.1 within 192 bp, and to r (2) = 0.05 within 451 bp. Population structure analysis using unlinked SNPs demonstrated the presence of two main distinct clusters representing western and eastern parts of the loblolly pine range included in our sample of trees.

Conclusions: The obtained results demonstrated the efficiency of exome capture for genotyping species such as loblolly pine with a large and complex genome. The highly diverse genetic variation reported in this study will be a valuable resource for future genetic and genomic research in loblolly pine.

Keywords: Exome sequence capture; Genotyping by sequencing; Linkage disequilibrium; Loblolly pine; Population structure; SNPs; Target enrichment.

Fig. 1

The counties of origin of the maternal trees colored by states. This map shows the sampling sites of the 362 out of 375 maternal parents of the ADEPT2 population used in this study

Fig. 2

Relationship between reads and capture target bases. a Relationship between reads and numbers of covered capture target bases. The numbers of captured target nucleotide bases are plotted against total number of sequence reads obtained in 375 trees from exome capture sequencing. The linear regression coefficient (r ²) is 0.23 (P < 0.001). b Distribution of on-target coverage ≥ 10x depth across the 375 trees. The numbers of capture target bases with a coverage depth of ten or greater sequence reads per target are plotted against the total number of sequence reads. The relationship seemed approximately linear within a limited range of the total number of reads at 37–55 million. c Distribution of mean coverage depth across the 375 trees. The mean coverage depth is plotted against the total number of sequence reads. The linear regression coefficients (r ²) was significant (P < 0.001) and equalled 0.72

Fig. 3

Cumulative distribution of coverage depth of captured target bases in 375 trees. Each line represents a single tree

Fig. 4

F _IS (left) and heterozygosity (right) distributions among 375 trees

Fig. 5

Linkage disequilibrium decay plot for 375 trees based on 972,720 SNP markers. Pairwise linkage disequilibrium coefficients (r ²) calculated for all 375 trees were plotted against the physical distances (bp) between all pairs of SNPs within the same scaffolds (left) and between pairs of SNPs within the same scaffolds located within 4000 bp (right). The trendlines of the nonlinear regressions (r ²) against physical distance between the SNPs are indicated in red

Fig. 6

Summarized admixture proportion distributions for K = 2 and K = 7. a & b Summarized admixture proportions plotted on the map. Each pie chart is partitioned via summarized population assignments inferred by fastStructure. c & d Individual tree admixture proportion distributions. The trees are aligned on the x-axis according to the longitude from west to east