Topic ID No.: 2026-G10
Title of research topic:
Using data science approaches and deep learning for characterizing the genetic and phenotypic diversity of quinoa and amaranth as grain and vegetable crop
Description:
Combining data science and plant breeding to develop improved crops to achieve a more sustainable and secure food system for healthy nutrition. The crops of interest are Quinoa and Amaranth which both have potential for cultivation in South Africa.
Target region or country (if applicable):
Topic is not yet linked to a specific country, but South Africa and/or Malawi are feasible options.
Topic background information / scientific relevance:
World food security depends on a few major crops. However, there is a large number of minor or underutilized crops which have a great potential to contribute to future food security and diversified agricultural systems. Two of these crops are amaranth and quinoa that have a potential for cultivation in South Africa. In contrast to major crops, there are fewer public data available that can be leveraged to develop locally adapted varieties using plant breeding. By using data science approaches of public data, however, it is possible to develop analysis and approaches that allow to identify genotypes that can be used as parents for crosses to establish breeding programs. Furthermore, it has been shown that both quinoa and emerald show a substantial heterosis and therefore by investigating the genetic basis of heterosis and identifying suitable parents for breeding programs, it is possible to tailor breeding schemes and selection in variety development. Amaranth and quinoa have a substantial potential for cultivation in South Africa and are and especially amaranth is already widely distributed as vegetable crop in this region. Therefore, it is of great interest to investigate the genetic basis of adaptation and heterosis of these crops as a prerequisite to develop improved varieties for cultivation in South Africa. Both greenhouse and field trials with a diversity of genotypes can be used in combination with data science and deep learning to investigate agronomically important traits such as grain or leaf yield and use them in subsequent genetic analyses.
Research objectives:
- Use data science approaches to assemble public data (genomic, phenotypic and climate) on amaranth and/or quinoa to select suitable accessions as parents in breeding programs
- Evaluate suitable parents in field trials in South Africa an/or Malawi, and in Europe
- Establish crosses from best parents and evaluate heterosis
- Implement and apply deep learning approaches using image analysis for phenotyping useful plant traits
- Genotype material and develop apply genetic mapping and genomic prediction
Required skills and qualifications of the applicant:
- A very good background in plant genetics or plant breeding
- Prior experience in data science and computational analysis (e.g., Linux, R, Python)
- Experiments on master level in experimental working with plants (greenhouse and/or field trials)
- Independent working style, but also ability to work in a team, affinity to computer-based data analysis, excellent good organization and communication skills
Contact person and institute in charge:
Prof. Dr. Karl Schmid. University of Hohenheim. Crop Plant Biodiversity and Breeding Informatics.