1.       Specific Aims

We propose to develop a generalized early warning system for citrus crop responses to pathogen infections by (a) robust detection of emitted volatile organics using training/test data sets under both laboratory and field conditions and by (b) constructing models and knowledge repositories for citrus pathogen defense mechanisms 

This dual strategy of developing a diagnostic system and integration of advanced computational tools validating diagnostic signals by mapping citrus regulatory systems could be used to activate a first line of defense against potentially industry-eliminating diseases. Diagnosis of citrus infections will be based on a reagentless detection platform consisting of a miniature solid-state chemical sensor, which will be modified into a hand-held low-power device suitable for field detection. A hand held differential mobility spectrometer (DMS) will be trained using increasing degrees of complexity of volatile emission patterns in a series of laboratory-to-field study designs. Initial pattern recognition models will be worked out from plant host response to infection under standardized laboratory conditions to yield optimal accuracy of diagnosis under a workflow of supervised learning algorithms (SVMs, trees, ANN, DFA, PLS and others). Consensus models consisting of the most robust DMS features will be built to result in most reliable biomarkers which will subsequently tested under field conditions. Simultaneously, volatile analysis will be performed using adsorbent trap columns which are used in connection with a gas chromatograph/time of flight mass spectrometer. Automatic peak picking and mass spectral deconvolution will feed into a multi-layered filtering algorithm in order to generate a customized volatile metabolite database. Spectra will be annotated to features detected by the DMS instrument in order to validate biomarkers. In addition, spectra will be associated with compound structures and biochemical pathways. In a third stream of information, gene microarray results from infected citrus plants will be mapped against three sequenced plant genomes in order to annotate citrus genes, yielding the first citrus gene annotation repository. Differential expression of genes will be interpreted by constructing a literature network repository using the Agilent freeware query tool in conjunction with Cytoscape 2.1. These efforts on generating repositories for citrus response pathways comprising literature knowledge, gene regulation and metabolic pathways will be combined to a unique pathogen defense knowledgebase for these crops that will validate and improve utility of diagnostic features derived from DMS pattern recognition. Figure 1 highlights a broad overview how the different sub-aims are tied together..

The heart of the project is the development of robust disease recognition features by working out data workflows and statistical tools from de-noising raw data, feature extraction and combination of supervised data models using data generated by the DMS systems, also called  generalized reagentless sensor system (GRSS). The GRSS will be developed for the real-time evaluation of internal tree and fruit physiology by sampling volatile organic compounds (VOCs) liberated into the air. Specific patterns of VOCs will be used to predict citrus plant/fruit health status using independent training and test sample data sets. Individual VOCs with high predictive power will be identified by off-line sampling and mass spectrometry analysis. The biological basis for release of these citrus VOCs will be validated by metabolite and gene profiling.  Together, a computational knowledgebase will be formed to inform automated recognition of citrus diseases, and to allow researchers to understand and monitor plant infection responses on a systems biology level by integrating gene and metabolite data using plant genome and literature-based pathway tools. Over three years, our team will identify the specific responses in citrus trees to abiotic (drought) and biotic stressors (pathogen/pests) - and develop host-response biomarkers using a bench top system that will be used to validate experimental plant material exposed to drought stress and natural infections, emphasizing the serious citrus pathogen, the Citrus tristeza virus (CTV). We plan to select a small and reliable set of validated markers to be incorporated into a hand held device that can be tested in the lab. Subsequently, a mobile prototype will be evaluated in the field and in laboratories settings at strategic locations. 

We have assembled a unique multi-disciplinary team of scientists from UC Davis to conduct this research, including researchers from the Genome Center, Plant Sciences, and Mechanical & Aeronautical Engineering Departments. The outcome will be development of a GRSS that can be used by growers, researchers, and regulators to monitor plant health status and validate prevention and treatment strategies. Finally, such a system could be integrated into robotics and other precision agriculture systems under development.

Goal:  Validate in-field diagnostic features of internal physiological disorders and pests in citrus orchards by informing pathogen response pathways using regulation of gene and metabolite expression.

Specific Aim 1: Derive statistically robust patterns of volatile emissions from pathogen-infected citrus crops using a hand-held and a benchtop DMS sensor system under laboratory and under diverse field conditions.

Specific Aim 2: Validate the prediction power of the DMS sensor system by constructing a GC-TOF mass spectrometry database for volatile metabolites from healthy and diseased citrus phenotypes. Annotate DMS features by chemical structures and metabolic pathways using this volatile compound database.

Specific Aim 3: Generate a citrus pathogen response repository by integrating volatile emission patterns with gene regulatory networks which are compared to a literature-query based background knowledgebase. Annotate citrus genes in a comparative genomics approach by mapping to known plant genomes (Arabidopsis thaliana, poplar and Medicago truncatula).