Joining the omics family, Phenomics is expected to play a key role in plant biology research. PhenomicsWorld is a project that studies the complex phenotypes of various species. Among the many phenotypes of interest, we initially focus on visually-observed phenotypes, which are difficult to articulate or be quantitatively/qualitatively characterized by plant geneticists but are typically suitable for capture through imagery. This project will provide the genomics community with a computational framework to study complex phenotypes using techniques such as visual content management, semantic modeling, knowledge sharing, and ontology customization.
Visual Phenotype Database System for Maize
We are currently working with the maize community to study mutants and diseases. To streamline the process of plant morphology, ecology, and phytochemistry research, a novel phenotype search engine, called VPhenoDBS, has been developed. This engine is unique in the sense that it provides query by phenotype images, query by semantics in mutant and disease descriptions, and other customized complex query methods to assist plant researchers studying the underlying effects of genetics and environmental factors on physiology. The currently available search methods are:
- Query by Image Example: Use a phenotype image to search for visually-similar phenotypes.
- Query by Semantics: Use semantic terms to search for matching phenotype images. The terms are mathematically modeled based on image content, not metadata or captions.
- Query by Text Annotation: A more traditional free-text search engine on the mutants in MaizeGDB, except the Plant and Gene Ontologies are leveraged to increase the accuracy of the search.
Interested in Collaborating?
This computational framework is highly configurable to other plants. We are actively looking for collaborators in other plant communities. If you're interested, please visit our Contact page
Here we supply a dataset containing images and semantic labels. The images are of leaves of various lesion mimic mutants and were taken in one of our fields here at MU. The dataset consists of 310 leaf images. The various genotypes present in the dataset contain a variety of lesions, including variations in lesion color, shape, size, and distribution. Accompanying the images is a CSV file containing semantic labels (e.g. image X has extensive brown lesions). It is our hope that this dataset will be useful for others studying phenotypes from imagery and desiring to segment and quantify lesion burden.
Click here to download the ISO file containing the dataset (2.2 GB).