Genome Wide Association Study and Genomic Selection of Amino Acid Concentrations in Soybean Seeds

Abstract

Soybean is a major source of protein for human consumption and animal feed. Releasing new cultivars with high nutritional value is one of the major goals in soybean breeding. To achieve this goal, genome-wide association studies of seed amino acid contents were conducted based on 249 soybean accessions from China, US, Japan, and South Korea. The accessions were evaluated for 15 amino acids and genotyped by sequencing. Significant genetic variation was observed for amino acids among the accessions. Among the 231 single nucleotide polymorphisms (SNPs) significantly associated with variations in amino acid contents, fifteen SNPs localized near 14 candidate genes involving in amino acid metabolism. The amino acids were classified into two groups with five in one group and seven amino acids in the other. Correlation coefficients among the amino acids within each group were high and positive, but the correlation coefficients of amino acids between the two groups were negative. Twenty-five SNP markers associated with multiple amino acids can be used to simultaneously improve multi-amino acid concentration in soybean. Genomic selection analysis of amino acid concentration showed that selection efficiency of amino acids based on the markers significantly associated with all 15 amino acids was higher than that based on random markers or markers only associated with individual amino acid. The identified markers could facilitate selection of soybean varieties with improved seed quality.

Keywords: Glycine max; amino acid concentration; genome-wide association study; genomic selection; genotyping by sequencing; single nucleotide polymorphism.

Figure 1

The maximum likelihood tree of the 20 soybean germplasm accessions that ranked in the top three for at least one amino acid concentration among the 249 soybean accessions.

Figure 2

Structure analysis: (A) Delta K values for different numbers of populations (K) from the STRUCTURE analysis, x-axes shows different numbers of populations (K), y-axes shows Delta K values for different numbers of populations (K). (B) Classification of 249 accessions into six sub-populations using STRUCTURE version 2.3.4, where the x-axis shows accessions, and the y-axis shows the probability (from 0 to 1) of each accession belong to sub-population (Q = K) membership. The membership of each accessions belonging to sub-populations is indicated by different colors (Q1, red; Q2, green; Q3, blue; Q4, yellow; Q5, pink; and Q6: cyan). (C) Maximum Likelihood (ML) tree of the 249 accessions drawn in MEGA 6. The color code for each subpopulation is the same as that in the (B and C).

Figure 3

The QQ plot between the expected LOD (-log(P-value)) value and the estimated LOD (log(P-value)) value of amino acid Ile based on 23,279 SNPs as an example (all 15 QQ-plot for the 15 amino acids showed in Supplementary Figure 3).

Figure 4

The correlation coefficient (r) among 15 amino acids between the observed values (each amino acid concentration) and the GEBVs predicted from the 231 SNP markers using RR-BLUP in rrBLUP software.

Figure 5

The average correlation coefficient (r) among 15 amino acids between the observed values (each amino acid concentration) and the GEBVs predicted in both training set and validation set from the 231 SNP markers using cBLUP method in GAPIT.