Authors
Yingwei Feng, Tingxi Yu, Congcong Shen, Kunhui He, Shang Gao, Zhiliang Zhang, Zhangping Huang, Fan Zhang, Jianlong Xu, Jiankang Wang, Wensheng Wang, Huihui Li
Published in
Molecular plant. Nov 19, 2026. Epub Nov 19, 2026.
Abstract
Accurate prediction of complex agronomic traits poses a long-standing challenge for hybrid crop breeding, which may be resolved by integrating multi-omics data with shared-parent relationship information among hybrids. To accelerate variety development, we constructed the DNNGP-Pro platform for predictive modeling of complex hybrid traits in large-scale populations. In preliminary steps, the pipeline extracts local features from omics data via convolutional neural networks, then employs graph convolutional networks to integrate these single-omics features with hybrid shared-parent graph (HSPG), and captures long-range dependencies across multi-omics features using a Transformer module. Exhibiting comparable performance to DNNGP model on single-omics data, incorporating the HSPG with individual genomic (G), metabolomic (M), or transcriptomic (T) datasets respectively improved prediction accuracy by 2.98%, 5.54%, and 4.75%, for 13 traits across four hybrid populations. Dual-omics models (MT, GT, or GM) demonstrated 5.33%-9.23% higher accuracy than single-omics models across 10 traits, whereas the incorporating all three omics data (GMT) with the HSPG enhanced accuracy by 2.95%, 1.71%, and 4.39%, respectively. To demonstrate DNNGP-Pro application for breeding high-yield rice varieties, we conducted trait prediction in the Rice26040 population, training on 1,040 screened hybrids and validating predicted candidates through multi-location trials. The model identified the superior indica hybrid, Chunliangyou825, within this massive candidate pool, which demonstrated 23.62% and 8.01% increases in grain yield in Hangzhou, and 7.70% and 16.15% higher yields in Pingxiang, over control varieties Tianyouhuazhan and Fengliangyou4, respectively, which we verified through regional testing and submitted for variety certification. The present study demonstrates the prediction power and analytical versatility of DNNGP-Pro in accelerating hybrid development through large-scale complex trait prediction from multi-omics data and shared-parent relationship information among hybrids.
PMID:
43059179
Bibliographic data and abstract were imported from PubMed on 20 Nov 2026.
Read full publication at:
Please sign in
to see all details.
Advertisement
Stats
- Recommendations n/a n/a positive of 0 vote(s)
- Views 1
- Comments 0