Genomic dissection and genomic prediction of essential amino acid content in cottonseed identifies candidate genes and elite germplasm for nutritional quality improvement

The amino acid composition of cottonseed is a critical determinant of its protein quality; however, the utility of genomic prediction for improving the essential amino acid content remains insufficiently established. In this study, 259 upland cotton ( Gossypium hirsutum L.) germplasm accessions were evaluated for 18 cottonseed amino acids across five environments by integrating best linear unbiased prediction (BLUP), genome-wide association study (GWAS), haplotype analysis, candidate gene annotation, and genomic selection. The 18 amino acid traits exhibited moderate model-based broad-sense heritability (H² = 0.377–0.588), while the first five principal components collectively explained 60.02% of the phenotypic variations. Focusing on the nutritionally limiting amino acids lysine (Lys), methionine (Met), and threonine (Thr), 357 significant SNPs were detected, of which 57 were consistently identified across environments or BLUP-based analyses and consolidated into seven stable candidate intervals. Haplotype analysis identified favorable haplotypes associated with increased Lys, Met, and Thr levels. The integration of nonsynonymous variant screening, GO/KEGG enrichment analysis, homologous functional annotation, and expression profiling prioritized 25 candidate genes, including the transporter-related genes Gh_D09G006500, Gh_A08G006200, Gh_A09G006600, and Gh_A09G007200 . Genomic selection showed moderate prediction accuracies of 0.511, 0.531, and 0.637 for Lys, Met, and Thr, respectively, and identified candidate germplasm accessions with favorable combined genomic breeding values. These findings provide stable genomic regions, prioritized candidate genes, and germplasm resources for the targeted improvement of limiting essential amino acids and nutritional quality of cottonseed meal.