The field of plant phenomics seeks to quantify the multi-dimensional basis of plant phenotypes. Phenomics has traditionally encompassed physiology and development with screening for large effect genetic mutations. The use of phenomics for population-wide genomic analysis of complex quantitative traits has been limited previously due to the cost of sequencing, complex analysis, and technical limitations to quantifying individual phenotypes from hundreds of plants. Today, high throughput phenotyping (HTP) is a newly emergent field where individual phenotypes are quantified in large populations of plants (Furbank and Tester, 2011). This can be combined with population-wide genotyping to dissect the genetic basis of traits of interest by using genome wide association studies (GWAS) to precisely identify which alleles are responsible (Li et al, 2010). Together HTP with GWAS have the potential to revolutionize the rate of trait discovery and our understanding of the genetic basis of key plant traits such as growth and yield. However many technical challenges remain. Although genomic analysis has advanced significantly in recent years due to plummeting sequencing costs, GWAS has yet to be exploited for multi-dimensional phenomic analysis because implementation requires technically complex linking of hardware and software systems. The Borevitz Lab seeks to bring HTP and GWAS to a wider audience by creating a user friendly “seeds to trait” data pipeline that will reduce the barrier to entry for new users. Our goal is to create an open source pipeline for trait capture that will enable any lab to carry out these types of studies even with modest equipment and technical background.
The current bottlenecks for widespread application are
- lack of standards for phenotyping protocols
- lack of automation of the capture and processing of image data into quantitative values
- lack of standardization and simplification of GWAS analysis and visualization.
Our work will integrate quantitative genetic analysis into phenomics, permitting A) the distinction to be made between observable phenotypes and heritable genetic traits, B) the determination of the genetic architecture of these traits, and C) the discovery of how these traits interact with their environment.
We are working to streamline the required hardware and software tools for image acquisition, data extraction, and analysis (Figure 1). This backend software pipeline will be complemented by an online user interface that will allow researchers to rapidly design, schedule, analyze, and visualize growth chamber experiments to identify heritable traits of interest from 300 plants per trial. Time-series image data will be available online in an interactive system in real-time. The interface will provide users the tools to rerun particular analyses based on the initial results (e.g. change thresholding values for colour detection) to fine tune detection of cryptic phenotypes.
HTP requirements vary widely by plant species, growth conditions, and by the physiology being examined. To make our system widely useful, we are employing a modular design using open source and/or open standards. The idea behind this approach is that users may collect image data using many cameras and output formats, but the basic experimental flow is one in which images are processed to yield quantitative phenotypes, then these results are paired with genetic data for GWAS analysis to yield trait and trait loci information.
Figure 1. Overview of the data pipeline.