Dr Xavier Sirault1

Dr Bob Furbank1

1: CSIRO Plant Industry, Black MountainCnr Clunies Ross St & Barry DriveCanberra, ACT 2601

Xavier.sirault@csiro.au

Novel High Resolution tools at the HRPPC

An Ontology-centric Architecture for ExtensibleScientic Data Management Systems

Gavin Kennedy1,2

Dr Yuan-Fang Li3

2: School of ITEE, University of Queensland, St Lucia, QLD

3: Clayton School of IT, Monash University, Clayton, VIC

Gavin.kennedy@csiro.au

What is Plant Phenomics?

Phenome = Genome X Environment

Genomics is accelerating gene discovery but how do we capitalise on these data sets to establish gene function and development of new genotypes for agriculture?

High throughput and high resolution analysis capacity now the factor limiting discovery of new traits and varieties

“ In the next 50 years we must produce more foodthan we have consumed in the history of mankind”Megan Clarke, CSIRO CEO 2009

Phenomics from the Leaf to the Field

Imagine a plant breeder walking his trials logging plant performance distributed sensors with his mobile phone or logging on to Phenonet from home to view his wheat in real time

HRPPC: Canberra node of the Australian Plant Phenomics Facility

Infrastructure: 1500 m2 lab space 245 m2 greenhouse 260 m2 growth

cabinetsAnalytical tools packaged in:

1- Model Plant Module (HTP) 2- Crop-Plant Shoot Module (MTP) 3- Crop-Plant Root Module (MTP) 4- Crop-Plant Field Module (HTP)

Brachypodium distachyon

Arabidopsis thaliana

Gossypium species

Triticum and Hordeum species,Vigna unguiculata (cowpea),Cicer arietinum (chickpea),Zea mays (maize),Sorghum bicolor, …

Role Deep phenotyping Development of next generation tools to probe

plant function and performance (come and see us)

Far Infrared imaging• Canopy / leaf temperature• Water use / salt tolerance

Capitalising on new imaging technologies

Visible imaging• Plant area, biomass, structure• Senescence, relative chlorophyll

content, pathogenic lesions

Near IR imaging• Tissue water content• Soil water content

Chlorophyll Fluorescence imaging

• Physiological state of photosynthetic machinery

FTIR Imaging Spectroscopy / Hyperspectral imaging

• Cellular localisation of metabolites (sugars, protein, aromatics)

• Carbohydrates, pigments and proteins

Plant FunctionPlant Morphology

• Light Detection and Ranging (LiDAR)

• Micro-bolometer sensors (Far-Infrared)

• 4-CCD line scanner (NIR and visible split)

PlantScan: next generation phenotyping platform for n-dimensional

Models

Addressing issues with fluorescence and environmental

control

Automated features extraction and quantification of n-dimensional models

Jurgen Fripp CSIRO ICT E-Health Brisbane

Automated segmentation – extracted stem Bounding box extraction and Delauney triangulation for convex 3D hull

Height and total volume extraction

Volume over time

Sirault, Fripp and Furbank (in preparation)

An integrated phenotyping platform for Model Plants

• PAM Fluorescence imaging• Far Infrared imaging• Visible imaging for growth• Climate controlled in equilibration

chamber and imaging chambers

2500 plants per day

Applications:• 1001 genomes project - 65 re-sequenced Arabidopsis thaliana ecotypes under

analysis - with Detlef Weigel• USDA Brachypodium distachyon project

www.phenonet.com

Distributed Sensor Network for Phenomics

Measure and log range of environmental factors on field trials.

Zigby wireless transmitters: Thermopile Temp Sensor Humidity Ambient Temp Soil Moisture

Imaging: Estimate biomass; greeness index for fertilization; detect flowering; estimate yield.Imaging constrained: Develop smarter portable platforms.

OntologiesOntologies are a set of formalised terms that allow us to represent

knowledge about concepts and relationships in a domain.

Annotating with ontologies means describing a domain object or process.

Modelling with ontologies means classifying a domain object or process, and its relationship to other domain concepts.

This image shows the wheat plant on the left has increased “salt tolerance (TO:0006001)”

OBI:0000050 : “platform”“A platform is an object_aggregate that is the set of instruments and software needed to perform a process. “

OntologiesEvolutionary

Changes in Domain, Model & Data

Expressed in OWL (& RDF Schema)Provides syntax & semantics - enables reasoningExpressivity vs decidabilityValidation via reasoning

Designed to be open & interoperableFacilitates sharing, reuse & Integration

Maturing technology stacksAPIs, reasoners, triple stores, query engines

TrayScan

PODD

PlantScan

Phenonet

Phenomobile

PODDData Stores

PODDMetadataRepository

Data Metadata

Data

Metadata

The Phenomics Ontology Driven Data repository

A research data and metadata repository.

Managing Phenomics Data from Multiple Heterogeneous High Volume High Resolution Data Generation Platforms

A methodology for managing and publishing research data outputs.

A semantic web data resource.

Putting the OD in PODD

Basics: Ontologies as domain models for research data

Model domain objects as ontological objectsBase ontology: domain independentPhenomics ontology: domain specific

Organizes data logicallyRepresented as metadata objectsParent-child relationshipReferential relationship

Drives all operations in the data lifecycle

Domain Concepts OWL Classes

Attributes and relations OWL Predicates

Domain Objects OWL Individuals

Comments, descriptions OWL Annotations

The PODD Ontology

Platform

Project

Project Plan Investigation Analysis Event Genotype

MaterialTreatmentMaterial

Container Data

Gene

Sequence

TreatmentObservation/Phenotype

Measurement

MeasurementParameter

Environment

SexArchive

Data

Design

PODD Architecture

Objects represented semanticallySemantics (metadata) captured in RDF

Repository operations on RDF:Ingestion, retrieval, update, query & search, export

Backend Object Management: Fedora Commons

Fedora objects mapped to Java objects for:Business Logic LayerInterface Layer

Future Work

Annotation ServicesOntological tagging of PODD objectsAnnotation tools, search/discovery tools, browsers, etc.

Virtual Laboratory EnvironmentSupport Phenome to Genome (and back) discovery processesAnalyse linkages across data resourcesWorkflows for statistical inferences & mathematical modelling.Visualisation toolsetc...

ResourcesPlant Phenomics Test Instance: http://poddtest.plantphenomics.org.au/

Plant Phenomics Production Instance: http://podd.plantphenomics.org.au/

Mouse Phenomics Production Instance: http://podd.australianphenomics.org.au

PODD Project Website: http://projects.arcs.org.au/trac/podd

Contact: Gavin.Kennedy@csiro.auPh: +61413 337 819

This work is part of a National eResearch Architecture Taskforce (NeAT) project, supported by the Australian National Data Service (ANDS) through the Education Investment Fund (EIF) Super Science Initiative, and the Australian Research Collaboration Service (ARCS) through the National Collaborative Research Infrastructure Strategy Program.

The Team

PODD Project ManagerGavin Kennedy

University of Queensland eResearch Lab:Faith Davies (Developer)Simon McNaughton (Developer)Jane Hunter (eResearch Lab Leader)

APPF/HRPCC/CSIROXavier Sirault (Science Leader, HRPPC)Xueqin Wang (Tester, Documentor)Bob Furbank (APPF HRPPC Leader)

APPF/Plant Accelerator/Uni of AdelaideBogdan Masznicz (Bioinformatician)Mark Tester (APPF TPA Leader)

APNPhilip Wu (Developer)Martin Hamilton (Developer)Adrienne McKenzie (APN Head of Network Services)

Monash Univesity Yuan-Fang Li (Designer)

NeATAndrew Treloar (Deputy Director ANDS)Paul Coddington (Projects Manager, ARCS)

ALADonald Hobern (Director, ALA)