AVATARS – Advanced Virtuality and AugmenTed Reality AppRoaches in Seeds to Seeds – is a BMBF-funded project with 5 years of funding started in June 2019.
Advances in omics´ analyses and high-throughput plant phenotyping offer unprecedented opportunities for crop improvement. However, the enormous increase in data volume and complexity bears tremendous challenges for data analysis, visualization, and interpretation. The AVATARS project will use novel approaches for data integration and visualization based on modeling and mixed reality:
- to make data more accessible,
- to select relevant information in the multidimensional data space at various levels of resolution, and
- to perform analyses efficiently.
AVATARS will be composed of a central scientific module and two flanking modules of project co-ordination and communication/education. The latter will establish ties with scientific target groups in fundamental and applied research, plant breeding, and computer vision and will reach out to students, science journalists, and the interested general public with an education-oriented work package.
Scientifically, AVATARS will focus on seed formation of Brassica napus. We aim to develop a time-resolved virtual 3D seed model based on high-resolution MRI, high throughput X-ray CT and histological data. The 3D seed model will be transferred into a VR/AR environment allowing to interactively experience transcriptome, proteome, and metabolome data. Furthermore, the virtual 3D seed model will include epigenetic information of three genotypes grown in controlled conditions, either beneficial or detrimental to seed formation. The integration of genetic and environmental factors allows determining their effect on seed traits. Novel deep learning algorithms gathering information from phenotypic, environmental, and genotypic data collected in field trials of up to 400 breeding lines will support the prediction of agronomic important seed traits.
AVATARS will use infrastructures established by DPPN and de.NBI and will create links to large datasets from previous projects (BMBF: BreedPatH, IRFFA, PROGReSs; DFG: PREDICT) via translation experiments´ thus supporting their efficient integration and exploitation.