Smart tools for flexible wheat use – EVAGRAIN

1- Capturing the diversity of French wheat.

Two harvest campaigns (2021 and 2022), each including around 150 samples, covered 39 production sites, 99 varieties, and all French quality classes (BP, BPS, BAF, BB, BAU). A complementary dataset of 11,000 breadmaking tests from ARVALIS positioned EVAGRAIN samples within the historical variability of French wheats.

2 - Ensuring experimental consistency.

All flours were produced using the same laboratory mill, and all breadmaking trials were conducted by a single expert baker.

3 -Structuring and harmonizing the data.

A dedicated ontology was developed to standardize the description of breadmaking tests and related knowledge (“French bread / EVAGRAIN”). It provides a robust foundation for future decision-support systems.

4 - Measuring reference technological criteria.

For each sample, 21 criteria commonly used in the cereal industry were analysed: protein content, test weight, moisture, gluten and associated markers, damaged starch, and farinograph and alveograph parameters (W, P, L, Ie, WA, etc.). Such a comprehensive dataset is rarely available at this scale and offers a unique basis for integrated quality analysis.

5 - In-depth biochemical characterization.

Flours were analysed for so-called “minor” components: lipids (fatty acids of total lipids and starch-bound lipids), pentosans (total and soluble), and a detailed profile of storage proteins (gliadins, glutenins). In total, 31 biochemical traits were measured.

6 - Combining multiple levels of statistical and modelling approaches.

– Classical statistical methods (multiple linear regressions) to identify causal links between biochemical composition and technological behaviour.

– Probabilistic methods (Bayesian networks) integrating expert knowledge and breadmaking scoring grids.

– Machine-learning approaches to predict breadmaking performance from analytical criteria.

1 - Data structuring and openness

A dedicated ontology enabled the harmonization of technological test descriptions and supported a data paper consolidating the full dataset from the 290 EVAGRAIN samples.

2 - Biochemical composition and technological performance

Analyses of the 290 samples representative of French wheat diversity confirmed the central role of biochemical composition in determining technological performance. Beyond classical criteria, soluble pentosans emerged as key determinants, influencing water absorption, aqueous-phase viscosity and dough rheology.

The relationships established between composition, technological measurements (alveograph, farinograph), and breadmaking performance enabled the development of a conceptual scheme incorporating dough extensibility at moulding (DE). The elasticity index (Ie) stood out as the most informative alveograph criterion.

3 - From breadmaking scores to “quality profiles”

Probabilistic modelling of the 11,000 ARVALIS tests, combined with expert baker input, revealed three main technological profiles: Extensible, Straight, and Tenacious.

These quality profiles provide a more informative and diagnostic view than the global AFNOR score, highlighting structural weaknesses in wheats.

4 - Predicting profiles from analytical data

Machine-learning methods predicted these profiles with good accuracy, with the elasticity index Ie emerging as the strongest predictor. However, neither the global AFNOR breadmaking score nor individual subscores could be reliably predicted. Incorporating biochemical composition (proteins, pentosans, lipids) did not significantly improve predictive performance.

5 - Operational tools for the cereal sector

Two tools were developed:

- PANIPRO, a software tool calculating quality profiles based on the AFNOR protocol and offering a standardized reading of dough behaviour.

- A probabilistic visualization tool (Bayesian network) to support the interpretation of breadmaking results.

Together, they form the foundations of a future, more advanced decision-support system.

6- Assessment of biscuit-making quality

A new test adapted to rotary-moulded biscuit technology was developed using a standardized recipe. It relies on criteria that differ from those of the CTCPA laminated biscuit test, which is highly selective with respect to dough density. This new approach allows a broader range of wheats to be used while confirming that high protein levels are detrimental to biscuit quality.

Soluble pentosans are emerging as a key criterion for assessing the technological potential of wheats, and their measurement by near-infrared spectroscopy now appears feasible.

Determining the dough’s “quality profile” also shows strong potential to guide varietal choices according to end uses.

The data-visualization tool for breadmaking tests has been well received, particularly for training bakers and mill operators.

Immediate follow-ups include:

– implementing NIR prediction of soluble pentosans;

– deploying decision-support tools to guide varietal selection in a context of climate change.

EVAGRAIN has strengthened the transfer of scientific and technological knowledge to stakeholders in the cereal supply chain, contributing to innovation and competitiveness in the sector.

Cereal grains are the most important renewable resource for human food and animal feed. About 55% of the 35 MT French wheat production is exported each year making France a major actor on the international market. However, French wheat has to face the production of other wheat growing countries whose agronomical practices favour low production prices and/or high protein content of grains, one of the main quality criteria of wheat.
EVAGRAIN focuses on the technological facet of the wheat quality defined as the ability to meet expectations for a given end-use. Measuring the technological quality is crucial for determining the market price, but besides the protein content, very few other criteria are actually used. Yet, wheat quality is complex, especially as agricultural trends change: i) climate change imposes increasing abiotic constraints on crops and ii) new sustainable agricultural practices arise from the market and societal demands. As a result the harvest quality and quantity get more heterogeneous which has significant adverse consequences on the agri-food chain, from storage to bakery product quality. Clearly a more robust and versatile evaluation system of grain quality is needed to answer the quality demand for a large range of uses, to anticipate more severe quality variations consecutive to global warming and to compete on the international market.

The ambition of EVAGRAIN is to design a Decision Support System (DSS) which can integrate knowledge about wheat quality and deliver plausible interpretations of quality tests results: i) for various end-uses in industry and ii) based on analytical data. A second objective is to explore innovative analytical quality tests. Finally, a third objective is to support knowledge transfer from cereal science and technology to economic actors of the cereal sector.

To reach these objectives the DSS will involve model-based assessment systems allowing comprehensive accounts of the dependences between the behavioural properties (protein aggregation capacity, dough visco-elasticity…) and the quality criteria (dough stickiness, bread loaf volume, biscuit colour…). The final DSS will integrate knowledge and data about grain and cereal products from different sources as database, literature, existing models, experts…Especially, research in cereal science has shown that beyond the content and nature of the proteins other grain components, such as lipids and pentosans, but also water status can deeply influence the technological behaviour of grains and the cereal product quality. The project will investigate these compounds allowing to establish possible relationships with protein behaviours and grain quality. This new knowledge will be integrated to the DSS to improve its performances. The final system will be implemented as a web-tool, usable by any actor of the cereal sector eager to assess the quality of wheat grain with three major outputs:

-The prediction of the quality of wheat with respect to end-use.
- An explicit account of the reasoning underlying the prediction of quality
- An assessment of the uncertainty of the outcomes.

EVAGRAIN is an interdisciplinary project that combines modelling approaches, experimental research and technological developments. A strong expertise on wheat quality is gathered from the institutional and industrial partners of the project, which will be reinforced by a scientific and technical advisory committee selected from VegepolysValley stakeholders. The food industry is facing a growing need for process optimization based on detailed resource characterization. This project will pave the way for the development of new standards to qualify wheat grain by promoting new assessment practices. EVAGRAIN's operational development will be limited to bread-making and biscuits, for which there are more data from the literature and the expertise of EVAGRAIN's partners.