Pulse Brain · Growing Health Evidence Index
Tier 3 — Observational / field trialPeer-reviewedConventional

Stable QTNs for soybean seed protein content support genomic selection and simulation-based molecular breeding

Xu Sun; Bo Hu; Fulin Shan; Wen-Xia Li; Hailong Ning

The Crop Journal · 2026

Read source ↗ All evidence

Summary

This study identified 66 stable QTNs associated with soybean seed protein content across three environments in a panel of 494 accessions, using genome-wide association analysis and the mrMLM method. Genomic selection models incorporating these QTNs achieved prediction accuracies of 0.65–0.75 within the training population and 0.41–0.47 in independent validation, whilst simulation-based molecular breeding identified 19 promising cross combinations for developing high-protein soybean varieties. The work demonstrates the translational utility of genomic discoveries for crop improvement, with the cross GR103 × GR324 identified as particularly versatile across multiple selection methodologies.

Regional applicability

The genomic resources and breeding methodologies reported are applicable globally, though the QTN identifications and prediction accuracies are specific to the germplasm panel and environments used (geographic location not stated in abstract). Transferability of prediction accuracy and QTN effects to other soybean germplasm and UK growing conditions would require validation but the methodology is broadly transferable to UK breeding programmes if UK soybean cultivation expands.

Key measures

Quantitative trait nucleotides (QTNs), genome-wide association study (GWAS), single-nucleotide polymorphism (SNP) sets, BLUP values, Genomic BLUP (GBLUP) prediction accuracy, machine learning model performance (10-fold cross-validation and independent validation), genotypic values of simulated F1 hybrids

Outcomes reported

The study identified 66 quantitative trait nucleotides (QTNs) associated with soybean seed protein content using genome-wide association analysis, and evaluated genomic selection models and simulation-based molecular breeding strategies to predict breeding potential. Genomic BLUP and machine learning models achieved prediction accuracies of 0.652–0.748 in cross-validation and 0.405–0.469 in independent validation, with simulated progeny showing significantly higher genotypic values than parental populations.

Theme
Farming systems, soils & land use
Subject
Cereals & grains
Study type
Research
Study design
Genome-wide association study with genomic selection validation and simulation-based breeding
Source type
Peer-reviewed study
Status
Published
System type
Arable cereals
DOI
10.1016/j.cj.2026.09.006
Catalogue ID
NRmv7gv47w-000

Topic tags

Pulse AI · ask about this record

Dig deeper with Pulse AI.

Ask about this record, its theme or its relevance to UK farming and policy. Pulse AI uses selected catalogue evidence and cites the sources it draws on.