Unlocking the potential of wheat grain heterogeneity using machine vision
Lead Research Organisation:
Rothamsted Research
Department Name: Sustainable Soils and Crops
Abstract
Abstracts are not currently available in GtR for all funded research. This is normally because the abstract was not required at the time of proposal submission, but may be because it included sensitive information such as personal details.
Technical Summary
This project unlocks the potential of wheat grain heterogeneity using machine vision. We will develop a novel single seed hyperspectral imaging technology (HSI) integrated with next generation machine learning and use this to develop phenotyping tools to improve uniformity of grain quality traits in UK wheat and implement a seed sorting solution for the UK food industry (bread, biscuit and malted wheat).
We have demonstrated the potential of HSI for the non-destructive prediction of total protein content in single wheat grains and quantified striking genetic variation of wheat grain protein uniformity in the Watkins wheat landrace panel. We will carry out GWAS analysis and identify novel marker-trait associations.The associated genes/loci identified in the GWAS will be further prioritized through transcriptomics analysis on contrasting lines to identify genes differentially expressed for grain uniformity loci. We will also carry out detailed physiological field studies on subsets of contrasting lines to quantify effects of phenology, ear architecture and canopy traits to help interpret the basis of the variance in homogeneity of the grain quality traits
We will validate the allelic variation of prioritized candidate genes through developing and evaluating NILs and RILs for target genetic loci. In addition our plant breeding partner (DSV) will phenotype breeding parents using HSI imaging to assess grain protein content uniformity and crosses will be selected and progressed to F5 yield and grain quality plots as a demonstration of wheat breeding for enhanced grain quality through single seed HSI screening. Finally together with our partners in the food industry (CBRI, LECO, NFI and ABM) we will demonstrate the benefits of highly controlled protein levels and ratios of protein subclasses by developing food products (bread, biscuit and malted wheat) at commercial scale and evaluating these products by flavour, sensory and consumer profiling
We have demonstrated the potential of HSI for the non-destructive prediction of total protein content in single wheat grains and quantified striking genetic variation of wheat grain protein uniformity in the Watkins wheat landrace panel. We will carry out GWAS analysis and identify novel marker-trait associations.The associated genes/loci identified in the GWAS will be further prioritized through transcriptomics analysis on contrasting lines to identify genes differentially expressed for grain uniformity loci. We will also carry out detailed physiological field studies on subsets of contrasting lines to quantify effects of phenology, ear architecture and canopy traits to help interpret the basis of the variance in homogeneity of the grain quality traits
We will validate the allelic variation of prioritized candidate genes through developing and evaluating NILs and RILs for target genetic loci. In addition our plant breeding partner (DSV) will phenotype breeding parents using HSI imaging to assess grain protein content uniformity and crosses will be selected and progressed to F5 yield and grain quality plots as a demonstration of wheat breeding for enhanced grain quality through single seed HSI screening. Finally together with our partners in the food industry (CBRI, LECO, NFI and ABM) we will demonstrate the benefits of highly controlled protein levels and ratios of protein subclasses by developing food products (bread, biscuit and malted wheat) at commercial scale and evaluating these products by flavour, sensory and consumer profiling
Organisations
People |
ORCID iD |
| Malcolm Hawkesford (Principal Investigator) |
| Description | The Project is progressing successfully with no significant deviations from the project plan. Deploying a single grain-based protein prediction model using hyperspectral imaging (HSI) and machine learning, we have analysed single wheat grains from two field trials in the UK at John Innes Centre and Rothamsted Research for the AE Watkins landrace collection and generated phenotypes for grain protein content (GPC) and GPC heterogeneity (the variation among grains within a batch of grains from a single plot). By using ~90 million single nucleotide polymorphisms through a genome-wide association model, we have identified promising QTL at chromosomes 6A, 6D and 7D of the Watkins landraces that showed significant allelic variation for GPC. Furthermore, results showed GPC heterogeneity is a heritable trait and a stable QTL at chromosome 6A contributed ~17% of the heritable variance' This project aimed to 1) Develop a novel single seed phenotyping tool based on hyperspectral imaging technology (HSI) integrated with next generation machine learning 2) Explain the determinism of the variance of uniformity of single seed grain quality parameters and explore a broad range of both known, and novel and exotic wheat genotypes for previously undefinable unique single seed traits, this will allow breeders to target previously unavailable grain quality uniformity traits, as well as speed selection from segregating populations. 3) Deploy the single grain HSI technology as a novel molecular breeding tool by determining key genes controlling single grain quality uniformity traits and validating the candidate genes by developing lines with contrasting expressing of the novel genes which we will test in field experiments. 4) Demonstrate the application of the single seed phenotyping tool as a sorting technology at laboratory and pilot production scale for wheat. This will demonstrate the ultimate value of the approach by producing exemplar food products (bread, biscuit and malted wheat). To date (part way through the project) we have: 1) Developed and validated the novel single seed phenotyping tool based on hyperspectral imaging technology (HSI). 2) Run field trials at University of Nottingham and Rothamsted Research in 2022-23 and 2023-24 examining novel and exotic wheat genotypes at 2 fertilizer N levels (high N and low N) which allowed characterisation of GPC (Grain Protein Content) and GPC heterogeneity alongside a range of other physiological traits. Data analysis further exploring the basis of GPC heterogeneity is ongoing. 3) We have applied this novel tool to single wheat grains from two field trials in the UK at John Innes Centre and Rothamsted Research for the AE Watkins landrace collection and generated phenotypes for grain protein content (GPC) and GPC heterogeneity (the variation among grains within a batch of grains from a single plot). By using ~90 million single nucleotide polymorphisms through a genome-wide association model, we have identified promising QTL that showed significant allelic variation for GPC and GPC heterogeneity and demonstrated its utility for future research. 4) We have demonstrated utility of the method for screening unique wheat blends for bread and biscuit exemplars and illustrated how single-seed screening can be used to track and enhance quality traits from field to product. |
| Exploitation Route | Too early to say. |
| Sectors | Agriculture Food and Drink |