What's happening at IRRI
Thursday, June 09, 2016, 01:15pm - 02:15pm
Physiological modeling of phenotype for rice improvement in the genomics era: A philosophy, examples, and proposals
Michael Dingkuhn (IRRI)
The biological sciences, including their application in breeding, are undergoing fundamental changes with great cognitive and practical opportunities. In the past, we tried to infer genetic control from phenotype variation alone (Mendel’s peas); now, we know the genotype better than the phenotype and must try to understand what happens between them. We want to predict phenotype from the genome, as well as phenotypic plasticity (adaptability) from genome × environment information. Models are the currency that mediates these three realms: environment, phenotype, and genome. But our crop models are rooted in physiological, not genetic, knowledge of phenotype. How should these evolve?
Distinct model applications in support of breeding are (1) predicting optimal ideotypes as breeding targets and (2) extracting “hidden” adaptation traits from phenomics × environment data to discover their genomic control through association studies. Thereby, (1) is a forward application and (2) is a reverse application called heuristics. Phenotype prediction (1) currently falls into two approaches: (1.1) physiology-based ideotype optimization, considered “mechanistic” by some; and (1.2) genome optimization, statistically predicting the best-performing genotype by comparing sequence variation with phenotype variation (e.g., GS models). Approaches (1.1) and (1.2) appear worlds apart but should gradually converge, for example, through machine learning.
Recent progress in crop model adaptation is presented, thus:
Approach (1.1): Using a physiological model (SAMARA, capable of simulating phenotypic plasticity), a virtual breeding exercise provided optimal trait combinations that would increase rice yield potential in current and future climate scenarios. It was guided by existing trait diversity (phenomics), but we cannot yet link the model’s parameters to genes.
Approach (2): A simpler physiological model was used in reverse mode to assist in phenotyping (RIDEV). This increased the power of genomewide association studies (GWAS) and led to the discovery of QTLs controlling thermal adaptation and acclimation, including probable epigenetic mechanisms.
We propose a new research thrust that pulls together model-derived ideotype blueprints, a wealth of available phenomics and genomics data for key traits, and genome editing technologies to “assemble” higher-yielding, climate-smart, and lodging-resistant rice varieties.
Havener Auditorium, International Rice Research Institute, Los Baños, 4030 Laguna, Philippines